social psychology you are what you...

51
1 YOU ARE WHAT YOU LIKE Social Psychology Empirical Modeling Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes are predictable from digital records of human behavior." Proceedings of the National Academy of Sciences 110.15 (2013): 5802-5805.

Upload: others

Post on 12-Aug-2020

4 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

1

YOU ARE WHAT YOU LIKE

Social Psychology

Empirical Modeling

Kosinski Michal David Stillwell and Thore Graepel Private traits and attributes are predictable from digital records of human behavior Proceedings of the National Academy of Sciences 11015 (2013) 5802-5805

2

New kinds of data

Social Sensing via RFID

3

Engagement and Exploration

Standing face-to-face

Physical distance

Hand gesture posture

Conversation patterns

Frequency of interruptions

4

Computational Social Science

The science that investigates social phenomena through the medium of computing and statistical data processing

5

3 Common Approaches

Macroscope Virtual Lab Empirical Modeling

6

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

7

WE ARE WHAT WE SAY

Linguistics

Schwartz H Andrew et al Personality gender and age in the language of social media The open-vocabulary approach PloS one 89 (2013) e73791

Macroscope

8

Dataset

700 million words phrases and topic instances collected from 75000 volunteersrsquo FB posts

Record usersrsquo personality (5-factor) gender and age

9

What Words Do You Use

male

female

10

How Old Are You

13 - 18

19 - 22

23 - 29

30 - 65

11

Personality Traits

Extraversion

Introversion

12

Topics Across 4 Age-groups

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal Achal Shah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

13

Warm and Negative Words

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal Achal Shah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

Usage of ldquoIrdquo amp ldquoWerdquo

Huge-volume data + simple analysis crystal clear language use patterns

15

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

16

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

17

Empirical Modeling

Traditional mathematical or computational modelingTends to rely on many often unrealistic assumptions

Not generally tested in detail against data

Result is proliferation of models that exist in parallel and are often incompatible with each other

New sourcesscales of data allow both to learntest models and also calibrate them

Observations Models Lab Field Observations

18

PREDICTION OF COUNTY-LEVEL HEART DISEASE MORTALITY

19

Prediction Accuracy

20

21

22

Language Use in Tweets

23

Social media opens up a new window of what humans actually feel and think

24

YOU ARE WHAT YOU LIKE

Social Psychology

Empirical Modeling

Kosinski Michal David Stillwell and Thore Graepel Private traits and attributes are predictable from digital records of human behavior Proceedings of the National Academy of Sciences 11015 (2013) 5802-5805

25

Personality Prediction

Personality traitsGender age relationship status friends

Sexual orientation ethnicity religion political inclination

Addictive substances (alcohol drugs cigarette) parental separation

IQ 5-Factor model satisfaction with Life

26

Data Collection

9939220 Likes (55814 unique ones) from 58466 Facebook volunteers

Sports

Music

Books

Restaurants

Popular websites

27

Ground truth

Political Inclination

Sexual Orientation

Democrat Republican

Democratic GOP (Grand Old Party)

Democratic Party Republican Party

Homosexual Heterosexual

1 0 1 0

28

Ground truth

5-Factor ModelOpenness

Conscientiousness

Extraversion

Agreeableness

Stability

29

Ground truth

Satisfaction with Life (SWL)

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 2: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

2

New kinds of data

Social Sensing via RFID

3

Engagement and Exploration

Standing face-to-face

Physical distance

Hand gesture posture

Conversation patterns

Frequency of interruptions

4

Computational Social Science

The science that investigates social phenomena through the medium of computing and statistical data processing

5

3 Common Approaches

Macroscope Virtual Lab Empirical Modeling

6

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

7

WE ARE WHAT WE SAY

Linguistics

Schwartz H Andrew et al Personality gender and age in the language of social media The open-vocabulary approach PloS one 89 (2013) e73791

Macroscope

8

Dataset

700 million words phrases and topic instances collected from 75000 volunteersrsquo FB posts

Record usersrsquo personality (5-factor) gender and age

9

What Words Do You Use

male

female

10

How Old Are You

13 - 18

19 - 22

23 - 29

30 - 65

11

Personality Traits

Extraversion

Introversion

12

Topics Across 4 Age-groups

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal Achal Shah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

13

Warm and Negative Words

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal Achal Shah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

Usage of ldquoIrdquo amp ldquoWerdquo

Huge-volume data + simple analysis crystal clear language use patterns

15

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

16

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

17

Empirical Modeling

Traditional mathematical or computational modelingTends to rely on many often unrealistic assumptions

Not generally tested in detail against data

Result is proliferation of models that exist in parallel and are often incompatible with each other

New sourcesscales of data allow both to learntest models and also calibrate them

Observations Models Lab Field Observations

18

PREDICTION OF COUNTY-LEVEL HEART DISEASE MORTALITY

19

Prediction Accuracy

20

21

22

Language Use in Tweets

23

Social media opens up a new window of what humans actually feel and think

24

YOU ARE WHAT YOU LIKE

Social Psychology

Empirical Modeling

Kosinski Michal David Stillwell and Thore Graepel Private traits and attributes are predictable from digital records of human behavior Proceedings of the National Academy of Sciences 11015 (2013) 5802-5805

25

Personality Prediction

Personality traitsGender age relationship status friends

Sexual orientation ethnicity religion political inclination

Addictive substances (alcohol drugs cigarette) parental separation

IQ 5-Factor model satisfaction with Life

26

Data Collection

9939220 Likes (55814 unique ones) from 58466 Facebook volunteers

Sports

Music

Books

Restaurants

Popular websites

27

Ground truth

Political Inclination

Sexual Orientation

Democrat Republican

Democratic GOP (Grand Old Party)

Democratic Party Republican Party

Homosexual Heterosexual

1 0 1 0

28

Ground truth

5-Factor ModelOpenness

Conscientiousness

Extraversion

Agreeableness

Stability

29

Ground truth

Satisfaction with Life (SWL)

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 3: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

3

Engagement and Exploration

Standing face-to-face

Physical distance

Hand gesture posture

Conversation patterns

Frequency of interruptions

4

Computational Social Science

The science that investigates social phenomena through the medium of computing and statistical data processing

5

3 Common Approaches

Macroscope Virtual Lab Empirical Modeling

6

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

7

WE ARE WHAT WE SAY

Linguistics

Schwartz H Andrew et al Personality gender and age in the language of social media The open-vocabulary approach PloS one 89 (2013) e73791

Macroscope

8

Dataset

700 million words phrases and topic instances collected from 75000 volunteersrsquo FB posts

Record usersrsquo personality (5-factor) gender and age

9

What Words Do You Use

male

female

10

How Old Are You

13 - 18

19 - 22

23 - 29

30 - 65

11

Personality Traits

Extraversion

Introversion

12

Topics Across 4 Age-groups

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal Achal Shah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

13

Warm and Negative Words

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal Achal Shah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

Usage of ldquoIrdquo amp ldquoWerdquo

Huge-volume data + simple analysis crystal clear language use patterns

15

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

16

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

17

Empirical Modeling

Traditional mathematical or computational modelingTends to rely on many often unrealistic assumptions

Not generally tested in detail against data

Result is proliferation of models that exist in parallel and are often incompatible with each other

New sourcesscales of data allow both to learntest models and also calibrate them

Observations Models Lab Field Observations

18

PREDICTION OF COUNTY-LEVEL HEART DISEASE MORTALITY

19

Prediction Accuracy

20

21

22

Language Use in Tweets

23

Social media opens up a new window of what humans actually feel and think

24

YOU ARE WHAT YOU LIKE

Social Psychology

Empirical Modeling

Kosinski Michal David Stillwell and Thore Graepel Private traits and attributes are predictable from digital records of human behavior Proceedings of the National Academy of Sciences 11015 (2013) 5802-5805

25

Personality Prediction

Personality traitsGender age relationship status friends

Sexual orientation ethnicity religion political inclination

Addictive substances (alcohol drugs cigarette) parental separation

IQ 5-Factor model satisfaction with Life

26

Data Collection

9939220 Likes (55814 unique ones) from 58466 Facebook volunteers

Sports

Music

Books

Restaurants

Popular websites

27

Ground truth

Political Inclination

Sexual Orientation

Democrat Republican

Democratic GOP (Grand Old Party)

Democratic Party Republican Party

Homosexual Heterosexual

1 0 1 0

28

Ground truth

5-Factor ModelOpenness

Conscientiousness

Extraversion

Agreeableness

Stability

29

Ground truth

Satisfaction with Life (SWL)

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 4: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

4

Computational Social Science

The science that investigates social phenomena through the medium of computing and statistical data processing

5

3 Common Approaches

Macroscope Virtual Lab Empirical Modeling

6

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

7

WE ARE WHAT WE SAY

Linguistics

Schwartz H Andrew et al Personality gender and age in the language of social media The open-vocabulary approach PloS one 89 (2013) e73791

Macroscope

8

Dataset

700 million words phrases and topic instances collected from 75000 volunteersrsquo FB posts

Record usersrsquo personality (5-factor) gender and age

9

What Words Do You Use

male

female

10

How Old Are You

13 - 18

19 - 22

23 - 29

30 - 65

11

Personality Traits

Extraversion

Introversion

12

Topics Across 4 Age-groups

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal Achal Shah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

13

Warm and Negative Words

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal Achal Shah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

Usage of ldquoIrdquo amp ldquoWerdquo

Huge-volume data + simple analysis crystal clear language use patterns

15

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

16

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

17

Empirical Modeling

Traditional mathematical or computational modelingTends to rely on many often unrealistic assumptions

Not generally tested in detail against data

Result is proliferation of models that exist in parallel and are often incompatible with each other

New sourcesscales of data allow both to learntest models and also calibrate them

Observations Models Lab Field Observations

18

PREDICTION OF COUNTY-LEVEL HEART DISEASE MORTALITY

19

Prediction Accuracy

20

21

22

Language Use in Tweets

23

Social media opens up a new window of what humans actually feel and think

24

YOU ARE WHAT YOU LIKE

Social Psychology

Empirical Modeling

Kosinski Michal David Stillwell and Thore Graepel Private traits and attributes are predictable from digital records of human behavior Proceedings of the National Academy of Sciences 11015 (2013) 5802-5805

25

Personality Prediction

Personality traitsGender age relationship status friends

Sexual orientation ethnicity religion political inclination

Addictive substances (alcohol drugs cigarette) parental separation

IQ 5-Factor model satisfaction with Life

26

Data Collection

9939220 Likes (55814 unique ones) from 58466 Facebook volunteers

Sports

Music

Books

Restaurants

Popular websites

27

Ground truth

Political Inclination

Sexual Orientation

Democrat Republican

Democratic GOP (Grand Old Party)

Democratic Party Republican Party

Homosexual Heterosexual

1 0 1 0

28

Ground truth

5-Factor ModelOpenness

Conscientiousness

Extraversion

Agreeableness

Stability

29

Ground truth

Satisfaction with Life (SWL)

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 5: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

5

3 Common Approaches

Macroscope Virtual Lab Empirical Modeling

6

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

7

WE ARE WHAT WE SAY

Linguistics

Schwartz H Andrew et al Personality gender and age in the language of social media The open-vocabulary approach PloS one 89 (2013) e73791

Macroscope

8

Dataset

700 million words phrases and topic instances collected from 75000 volunteersrsquo FB posts

Record usersrsquo personality (5-factor) gender and age

9

What Words Do You Use

male

female

10

How Old Are You

13 - 18

19 - 22

23 - 29

30 - 65

11

Personality Traits

Extraversion

Introversion

12

Topics Across 4 Age-groups

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal Achal Shah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

13

Warm and Negative Words

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal Achal Shah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

Usage of ldquoIrdquo amp ldquoWerdquo

Huge-volume data + simple analysis crystal clear language use patterns

15

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

16

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

17

Empirical Modeling

Traditional mathematical or computational modelingTends to rely on many often unrealistic assumptions

Not generally tested in detail against data

Result is proliferation of models that exist in parallel and are often incompatible with each other

New sourcesscales of data allow both to learntest models and also calibrate them

Observations Models Lab Field Observations

18

PREDICTION OF COUNTY-LEVEL HEART DISEASE MORTALITY

19

Prediction Accuracy

20

21

22

Language Use in Tweets

23

Social media opens up a new window of what humans actually feel and think

24

YOU ARE WHAT YOU LIKE

Social Psychology

Empirical Modeling

Kosinski Michal David Stillwell and Thore Graepel Private traits and attributes are predictable from digital records of human behavior Proceedings of the National Academy of Sciences 11015 (2013) 5802-5805

25

Personality Prediction

Personality traitsGender age relationship status friends

Sexual orientation ethnicity religion political inclination

Addictive substances (alcohol drugs cigarette) parental separation

IQ 5-Factor model satisfaction with Life

26

Data Collection

9939220 Likes (55814 unique ones) from 58466 Facebook volunteers

Sports

Music

Books

Restaurants

Popular websites

27

Ground truth

Political Inclination

Sexual Orientation

Democrat Republican

Democratic GOP (Grand Old Party)

Democratic Party Republican Party

Homosexual Heterosexual

1 0 1 0

28

Ground truth

5-Factor ModelOpenness

Conscientiousness

Extraversion

Agreeableness

Stability

29

Ground truth

Satisfaction with Life (SWL)

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 6: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

6

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

7

WE ARE WHAT WE SAY

Linguistics

Schwartz H Andrew et al Personality gender and age in the language of social media The open-vocabulary approach PloS one 89 (2013) e73791

Macroscope

8

Dataset

700 million words phrases and topic instances collected from 75000 volunteersrsquo FB posts

Record usersrsquo personality (5-factor) gender and age

9

What Words Do You Use

male

female

10

How Old Are You

13 - 18

19 - 22

23 - 29

30 - 65

11

Personality Traits

Extraversion

Introversion

12

Topics Across 4 Age-groups

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal Achal Shah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

13

Warm and Negative Words

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal Achal Shah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

Usage of ldquoIrdquo amp ldquoWerdquo

Huge-volume data + simple analysis crystal clear language use patterns

15

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

16

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

17

Empirical Modeling

Traditional mathematical or computational modelingTends to rely on many often unrealistic assumptions

Not generally tested in detail against data

Result is proliferation of models that exist in parallel and are often incompatible with each other

New sourcesscales of data allow both to learntest models and also calibrate them

Observations Models Lab Field Observations

18

PREDICTION OF COUNTY-LEVEL HEART DISEASE MORTALITY

19

Prediction Accuracy

20

21

22

Language Use in Tweets

23

Social media opens up a new window of what humans actually feel and think

24

YOU ARE WHAT YOU LIKE

Social Psychology

Empirical Modeling

Kosinski Michal David Stillwell and Thore Graepel Private traits and attributes are predictable from digital records of human behavior Proceedings of the National Academy of Sciences 11015 (2013) 5802-5805

25

Personality Prediction

Personality traitsGender age relationship status friends

Sexual orientation ethnicity religion political inclination

Addictive substances (alcohol drugs cigarette) parental separation

IQ 5-Factor model satisfaction with Life

26

Data Collection

9939220 Likes (55814 unique ones) from 58466 Facebook volunteers

Sports

Music

Books

Restaurants

Popular websites

27

Ground truth

Political Inclination

Sexual Orientation

Democrat Republican

Democratic GOP (Grand Old Party)

Democratic Party Republican Party

Homosexual Heterosexual

1 0 1 0

28

Ground truth

5-Factor ModelOpenness

Conscientiousness

Extraversion

Agreeableness

Stability

29

Ground truth

Satisfaction with Life (SWL)

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 7: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

7

WE ARE WHAT WE SAY

Linguistics

Schwartz H Andrew et al Personality gender and age in the language of social media The open-vocabulary approach PloS one 89 (2013) e73791

Macroscope

8

Dataset

700 million words phrases and topic instances collected from 75000 volunteersrsquo FB posts

Record usersrsquo personality (5-factor) gender and age

9

What Words Do You Use

male

female

10

How Old Are You

13 - 18

19 - 22

23 - 29

30 - 65

11

Personality Traits

Extraversion

Introversion

12

Topics Across 4 Age-groups

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal Achal Shah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

13

Warm and Negative Words

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal Achal Shah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

Usage of ldquoIrdquo amp ldquoWerdquo

Huge-volume data + simple analysis crystal clear language use patterns

15

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

16

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

17

Empirical Modeling

Traditional mathematical or computational modelingTends to rely on many often unrealistic assumptions

Not generally tested in detail against data

Result is proliferation of models that exist in parallel and are often incompatible with each other

New sourcesscales of data allow both to learntest models and also calibrate them

Observations Models Lab Field Observations

18

PREDICTION OF COUNTY-LEVEL HEART DISEASE MORTALITY

19

Prediction Accuracy

20

21

22

Language Use in Tweets

23

Social media opens up a new window of what humans actually feel and think

24

YOU ARE WHAT YOU LIKE

Social Psychology

Empirical Modeling

Kosinski Michal David Stillwell and Thore Graepel Private traits and attributes are predictable from digital records of human behavior Proceedings of the National Academy of Sciences 11015 (2013) 5802-5805

25

Personality Prediction

Personality traitsGender age relationship status friends

Sexual orientation ethnicity religion political inclination

Addictive substances (alcohol drugs cigarette) parental separation

IQ 5-Factor model satisfaction with Life

26

Data Collection

9939220 Likes (55814 unique ones) from 58466 Facebook volunteers

Sports

Music

Books

Restaurants

Popular websites

27

Ground truth

Political Inclination

Sexual Orientation

Democrat Republican

Democratic GOP (Grand Old Party)

Democratic Party Republican Party

Homosexual Heterosexual

1 0 1 0

28

Ground truth

5-Factor ModelOpenness

Conscientiousness

Extraversion

Agreeableness

Stability

29

Ground truth

Satisfaction with Life (SWL)

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 8: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

8

Dataset

700 million words phrases and topic instances collected from 75000 volunteersrsquo FB posts

Record usersrsquo personality (5-factor) gender and age

9

What Words Do You Use

male

female

10

How Old Are You

13 - 18

19 - 22

23 - 29

30 - 65

11

Personality Traits

Extraversion

Introversion

12

Topics Across 4 Age-groups

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal Achal Shah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

13

Warm and Negative Words

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal Achal Shah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

Usage of ldquoIrdquo amp ldquoWerdquo

Huge-volume data + simple analysis crystal clear language use patterns

15

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

16

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

17

Empirical Modeling

Traditional mathematical or computational modelingTends to rely on many often unrealistic assumptions

Not generally tested in detail against data

Result is proliferation of models that exist in parallel and are often incompatible with each other

New sourcesscales of data allow both to learntest models and also calibrate them

Observations Models Lab Field Observations

18

PREDICTION OF COUNTY-LEVEL HEART DISEASE MORTALITY

19

Prediction Accuracy

20

21

22

Language Use in Tweets

23

Social media opens up a new window of what humans actually feel and think

24

YOU ARE WHAT YOU LIKE

Social Psychology

Empirical Modeling

Kosinski Michal David Stillwell and Thore Graepel Private traits and attributes are predictable from digital records of human behavior Proceedings of the National Academy of Sciences 11015 (2013) 5802-5805

25

Personality Prediction

Personality traitsGender age relationship status friends

Sexual orientation ethnicity religion political inclination

Addictive substances (alcohol drugs cigarette) parental separation

IQ 5-Factor model satisfaction with Life

26

Data Collection

9939220 Likes (55814 unique ones) from 58466 Facebook volunteers

Sports

Music

Books

Restaurants

Popular websites

27

Ground truth

Political Inclination

Sexual Orientation

Democrat Republican

Democratic GOP (Grand Old Party)

Democratic Party Republican Party

Homosexual Heterosexual

1 0 1 0

28

Ground truth

5-Factor ModelOpenness

Conscientiousness

Extraversion

Agreeableness

Stability

29

Ground truth

Satisfaction with Life (SWL)

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 9: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

9

What Words Do You Use

male

female

10

How Old Are You

13 - 18

19 - 22

23 - 29

30 - 65

11

Personality Traits

Extraversion

Introversion

12

Topics Across 4 Age-groups

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal Achal Shah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

13

Warm and Negative Words

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal Achal Shah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

Usage of ldquoIrdquo amp ldquoWerdquo

Huge-volume data + simple analysis crystal clear language use patterns

15

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

16

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

17

Empirical Modeling

Traditional mathematical or computational modelingTends to rely on many often unrealistic assumptions

Not generally tested in detail against data

Result is proliferation of models that exist in parallel and are often incompatible with each other

New sourcesscales of data allow both to learntest models and also calibrate them

Observations Models Lab Field Observations

18

PREDICTION OF COUNTY-LEVEL HEART DISEASE MORTALITY

19

Prediction Accuracy

20

21

22

Language Use in Tweets

23

Social media opens up a new window of what humans actually feel and think

24

YOU ARE WHAT YOU LIKE

Social Psychology

Empirical Modeling

Kosinski Michal David Stillwell and Thore Graepel Private traits and attributes are predictable from digital records of human behavior Proceedings of the National Academy of Sciences 11015 (2013) 5802-5805

25

Personality Prediction

Personality traitsGender age relationship status friends

Sexual orientation ethnicity religion political inclination

Addictive substances (alcohol drugs cigarette) parental separation

IQ 5-Factor model satisfaction with Life

26

Data Collection

9939220 Likes (55814 unique ones) from 58466 Facebook volunteers

Sports

Music

Books

Restaurants

Popular websites

27

Ground truth

Political Inclination

Sexual Orientation

Democrat Republican

Democratic GOP (Grand Old Party)

Democratic Party Republican Party

Homosexual Heterosexual

1 0 1 0

28

Ground truth

5-Factor ModelOpenness

Conscientiousness

Extraversion

Agreeableness

Stability

29

Ground truth

Satisfaction with Life (SWL)

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 10: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

10

How Old Are You

13 - 18

19 - 22

23 - 29

30 - 65

11

Personality Traits

Extraversion

Introversion

12

Topics Across 4 Age-groups

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal Achal Shah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

13

Warm and Negative Words

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal Achal Shah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

Usage of ldquoIrdquo amp ldquoWerdquo

Huge-volume data + simple analysis crystal clear language use patterns

15

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

16

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

17

Empirical Modeling

Traditional mathematical or computational modelingTends to rely on many often unrealistic assumptions

Not generally tested in detail against data

Result is proliferation of models that exist in parallel and are often incompatible with each other

New sourcesscales of data allow both to learntest models and also calibrate them

Observations Models Lab Field Observations

18

PREDICTION OF COUNTY-LEVEL HEART DISEASE MORTALITY

19

Prediction Accuracy

20

21

22

Language Use in Tweets

23

Social media opens up a new window of what humans actually feel and think

24

YOU ARE WHAT YOU LIKE

Social Psychology

Empirical Modeling

Kosinski Michal David Stillwell and Thore Graepel Private traits and attributes are predictable from digital records of human behavior Proceedings of the National Academy of Sciences 11015 (2013) 5802-5805

25

Personality Prediction

Personality traitsGender age relationship status friends

Sexual orientation ethnicity religion political inclination

Addictive substances (alcohol drugs cigarette) parental separation

IQ 5-Factor model satisfaction with Life

26

Data Collection

9939220 Likes (55814 unique ones) from 58466 Facebook volunteers

Sports

Music

Books

Restaurants

Popular websites

27

Ground truth

Political Inclination

Sexual Orientation

Democrat Republican

Democratic GOP (Grand Old Party)

Democratic Party Republican Party

Homosexual Heterosexual

1 0 1 0

28

Ground truth

5-Factor ModelOpenness

Conscientiousness

Extraversion

Agreeableness

Stability

29

Ground truth

Satisfaction with Life (SWL)

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 11: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

11

Personality Traits

Extraversion

Introversion

12

Topics Across 4 Age-groups

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal Achal Shah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

13

Warm and Negative Words

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal Achal Shah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

Usage of ldquoIrdquo amp ldquoWerdquo

Huge-volume data + simple analysis crystal clear language use patterns

15

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

16

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

17

Empirical Modeling

Traditional mathematical or computational modelingTends to rely on many often unrealistic assumptions

Not generally tested in detail against data

Result is proliferation of models that exist in parallel and are often incompatible with each other

New sourcesscales of data allow both to learntest models and also calibrate them

Observations Models Lab Field Observations

18

PREDICTION OF COUNTY-LEVEL HEART DISEASE MORTALITY

19

Prediction Accuracy

20

21

22

Language Use in Tweets

23

Social media opens up a new window of what humans actually feel and think

24

YOU ARE WHAT YOU LIKE

Social Psychology

Empirical Modeling

Kosinski Michal David Stillwell and Thore Graepel Private traits and attributes are predictable from digital records of human behavior Proceedings of the National Academy of Sciences 11015 (2013) 5802-5805

25

Personality Prediction

Personality traitsGender age relationship status friends

Sexual orientation ethnicity religion political inclination

Addictive substances (alcohol drugs cigarette) parental separation

IQ 5-Factor model satisfaction with Life

26

Data Collection

9939220 Likes (55814 unique ones) from 58466 Facebook volunteers

Sports

Music

Books

Restaurants

Popular websites

27

Ground truth

Political Inclination

Sexual Orientation

Democrat Republican

Democratic GOP (Grand Old Party)

Democratic Party Republican Party

Homosexual Heterosexual

1 0 1 0

28

Ground truth

5-Factor ModelOpenness

Conscientiousness

Extraversion

Agreeableness

Stability

29

Ground truth

Satisfaction with Life (SWL)

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 12: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

12

Topics Across 4 Age-groups

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal Achal Shah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

13

Warm and Negative Words

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal Achal Shah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

Usage of ldquoIrdquo amp ldquoWerdquo

Huge-volume data + simple analysis crystal clear language use patterns

15

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

16

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

17

Empirical Modeling

Traditional mathematical or computational modelingTends to rely on many often unrealistic assumptions

Not generally tested in detail against data

Result is proliferation of models that exist in parallel and are often incompatible with each other

New sourcesscales of data allow both to learntest models and also calibrate them

Observations Models Lab Field Observations

18

PREDICTION OF COUNTY-LEVEL HEART DISEASE MORTALITY

19

Prediction Accuracy

20

21

22

Language Use in Tweets

23

Social media opens up a new window of what humans actually feel and think

24

YOU ARE WHAT YOU LIKE

Social Psychology

Empirical Modeling

Kosinski Michal David Stillwell and Thore Graepel Private traits and attributes are predictable from digital records of human behavior Proceedings of the National Academy of Sciences 11015 (2013) 5802-5805

25

Personality Prediction

Personality traitsGender age relationship status friends

Sexual orientation ethnicity religion political inclination

Addictive substances (alcohol drugs cigarette) parental separation

IQ 5-Factor model satisfaction with Life

26

Data Collection

9939220 Likes (55814 unique ones) from 58466 Facebook volunteers

Sports

Music

Books

Restaurants

Popular websites

27

Ground truth

Political Inclination

Sexual Orientation

Democrat Republican

Democratic GOP (Grand Old Party)

Democratic Party Republican Party

Homosexual Heterosexual

1 0 1 0

28

Ground truth

5-Factor ModelOpenness

Conscientiousness

Extraversion

Agreeableness

Stability

29

Ground truth

Satisfaction with Life (SWL)

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 13: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

13

Warm and Negative Words

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal Achal Shah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

Usage of ldquoIrdquo amp ldquoWerdquo

Huge-volume data + simple analysis crystal clear language use patterns

15

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

16

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

17

Empirical Modeling

Traditional mathematical or computational modelingTends to rely on many often unrealistic assumptions

Not generally tested in detail against data

Result is proliferation of models that exist in parallel and are often incompatible with each other

New sourcesscales of data allow both to learntest models and also calibrate them

Observations Models Lab Field Observations

18

PREDICTION OF COUNTY-LEVEL HEART DISEASE MORTALITY

19

Prediction Accuracy

20

21

22

Language Use in Tweets

23

Social media opens up a new window of what humans actually feel and think

24

YOU ARE WHAT YOU LIKE

Social Psychology

Empirical Modeling

Kosinski Michal David Stillwell and Thore Graepel Private traits and attributes are predictable from digital records of human behavior Proceedings of the National Academy of Sciences 11015 (2013) 5802-5805

25

Personality Prediction

Personality traitsGender age relationship status friends

Sexual orientation ethnicity religion political inclination

Addictive substances (alcohol drugs cigarette) parental separation

IQ 5-Factor model satisfaction with Life

26

Data Collection

9939220 Likes (55814 unique ones) from 58466 Facebook volunteers

Sports

Music

Books

Restaurants

Popular websites

27

Ground truth

Political Inclination

Sexual Orientation

Democrat Republican

Democratic GOP (Grand Old Party)

Democratic Party Republican Party

Homosexual Heterosexual

1 0 1 0

28

Ground truth

5-Factor ModelOpenness

Conscientiousness

Extraversion

Agreeableness

Stability

29

Ground truth

Satisfaction with Life (SWL)

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 14: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013orH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal Achal Shah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungar

Usage of ldquoIrdquo amp ldquoWerdquo

Huge-volume data + simple analysis crystal clear language use patterns

15

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

16

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

17

Empirical Modeling

Traditional mathematical or computational modelingTends to rely on many often unrealistic assumptions

Not generally tested in detail against data

Result is proliferation of models that exist in parallel and are often incompatible with each other

New sourcesscales of data allow both to learntest models and also calibrate them

Observations Models Lab Field Observations

18

PREDICTION OF COUNTY-LEVEL HEART DISEASE MORTALITY

19

Prediction Accuracy

20

21

22

Language Use in Tweets

23

Social media opens up a new window of what humans actually feel and think

24

YOU ARE WHAT YOU LIKE

Social Psychology

Empirical Modeling

Kosinski Michal David Stillwell and Thore Graepel Private traits and attributes are predictable from digital records of human behavior Proceedings of the National Academy of Sciences 11015 (2013) 5802-5805

25

Personality Prediction

Personality traitsGender age relationship status friends

Sexual orientation ethnicity religion political inclination

Addictive substances (alcohol drugs cigarette) parental separation

IQ 5-Factor model satisfaction with Life

26

Data Collection

9939220 Likes (55814 unique ones) from 58466 Facebook volunteers

Sports

Music

Books

Restaurants

Popular websites

27

Ground truth

Political Inclination

Sexual Orientation

Democrat Republican

Democratic GOP (Grand Old Party)

Democratic Party Republican Party

Homosexual Heterosexual

1 0 1 0

28

Ground truth

5-Factor ModelOpenness

Conscientiousness

Extraversion

Agreeableness

Stability

29

Ground truth

Satisfaction with Life (SWL)

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 15: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

15

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

16

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

17

Empirical Modeling

Traditional mathematical or computational modelingTends to rely on many often unrealistic assumptions

Not generally tested in detail against data

Result is proliferation of models that exist in parallel and are often incompatible with each other

New sourcesscales of data allow both to learntest models and also calibrate them

Observations Models Lab Field Observations

18

PREDICTION OF COUNTY-LEVEL HEART DISEASE MORTALITY

19

Prediction Accuracy

20

21

22

Language Use in Tweets

23

Social media opens up a new window of what humans actually feel and think

24

YOU ARE WHAT YOU LIKE

Social Psychology

Empirical Modeling

Kosinski Michal David Stillwell and Thore Graepel Private traits and attributes are predictable from digital records of human behavior Proceedings of the National Academy of Sciences 11015 (2013) 5802-5805

25

Personality Prediction

Personality traitsGender age relationship status friends

Sexual orientation ethnicity religion political inclination

Addictive substances (alcohol drugs cigarette) parental separation

IQ 5-Factor model satisfaction with Life

26

Data Collection

9939220 Likes (55814 unique ones) from 58466 Facebook volunteers

Sports

Music

Books

Restaurants

Popular websites

27

Ground truth

Political Inclination

Sexual Orientation

Democrat Republican

Democratic GOP (Grand Old Party)

Democratic Party Republican Party

Homosexual Heterosexual

1 0 1 0

28

Ground truth

5-Factor ModelOpenness

Conscientiousness

Extraversion

Agreeableness

Stability

29

Ground truth

Satisfaction with Life (SWL)

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 16: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

16

APPROACHES1 MACROSCOPE 2 VIRTUAL LAB3 EMPIRICAL MODELING

17

Empirical Modeling

Traditional mathematical or computational modelingTends to rely on many often unrealistic assumptions

Not generally tested in detail against data

Result is proliferation of models that exist in parallel and are often incompatible with each other

New sourcesscales of data allow both to learntest models and also calibrate them

Observations Models Lab Field Observations

18

PREDICTION OF COUNTY-LEVEL HEART DISEASE MORTALITY

19

Prediction Accuracy

20

21

22

Language Use in Tweets

23

Social media opens up a new window of what humans actually feel and think

24

YOU ARE WHAT YOU LIKE

Social Psychology

Empirical Modeling

Kosinski Michal David Stillwell and Thore Graepel Private traits and attributes are predictable from digital records of human behavior Proceedings of the National Academy of Sciences 11015 (2013) 5802-5805

25

Personality Prediction

Personality traitsGender age relationship status friends

Sexual orientation ethnicity religion political inclination

Addictive substances (alcohol drugs cigarette) parental separation

IQ 5-Factor model satisfaction with Life

26

Data Collection

9939220 Likes (55814 unique ones) from 58466 Facebook volunteers

Sports

Music

Books

Restaurants

Popular websites

27

Ground truth

Political Inclination

Sexual Orientation

Democrat Republican

Democratic GOP (Grand Old Party)

Democratic Party Republican Party

Homosexual Heterosexual

1 0 1 0

28

Ground truth

5-Factor ModelOpenness

Conscientiousness

Extraversion

Agreeableness

Stability

29

Ground truth

Satisfaction with Life (SWL)

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 17: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

17

Empirical Modeling

Traditional mathematical or computational modelingTends to rely on many often unrealistic assumptions

Not generally tested in detail against data

Result is proliferation of models that exist in parallel and are often incompatible with each other

New sourcesscales of data allow both to learntest models and also calibrate them

Observations Models Lab Field Observations

18

PREDICTION OF COUNTY-LEVEL HEART DISEASE MORTALITY

19

Prediction Accuracy

20

21

22

Language Use in Tweets

23

Social media opens up a new window of what humans actually feel and think

24

YOU ARE WHAT YOU LIKE

Social Psychology

Empirical Modeling

Kosinski Michal David Stillwell and Thore Graepel Private traits and attributes are predictable from digital records of human behavior Proceedings of the National Academy of Sciences 11015 (2013) 5802-5805

25

Personality Prediction

Personality traitsGender age relationship status friends

Sexual orientation ethnicity religion political inclination

Addictive substances (alcohol drugs cigarette) parental separation

IQ 5-Factor model satisfaction with Life

26

Data Collection

9939220 Likes (55814 unique ones) from 58466 Facebook volunteers

Sports

Music

Books

Restaurants

Popular websites

27

Ground truth

Political Inclination

Sexual Orientation

Democrat Republican

Democratic GOP (Grand Old Party)

Democratic Party Republican Party

Homosexual Heterosexual

1 0 1 0

28

Ground truth

5-Factor ModelOpenness

Conscientiousness

Extraversion

Agreeableness

Stability

29

Ground truth

Satisfaction with Life (SWL)

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 18: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

18

PREDICTION OF COUNTY-LEVEL HEART DISEASE MORTALITY

19

Prediction Accuracy

20

21

22

Language Use in Tweets

23

Social media opens up a new window of what humans actually feel and think

24

YOU ARE WHAT YOU LIKE

Social Psychology

Empirical Modeling

Kosinski Michal David Stillwell and Thore Graepel Private traits and attributes are predictable from digital records of human behavior Proceedings of the National Academy of Sciences 11015 (2013) 5802-5805

25

Personality Prediction

Personality traitsGender age relationship status friends

Sexual orientation ethnicity religion political inclination

Addictive substances (alcohol drugs cigarette) parental separation

IQ 5-Factor model satisfaction with Life

26

Data Collection

9939220 Likes (55814 unique ones) from 58466 Facebook volunteers

Sports

Music

Books

Restaurants

Popular websites

27

Ground truth

Political Inclination

Sexual Orientation

Democrat Republican

Democratic GOP (Grand Old Party)

Democratic Party Republican Party

Homosexual Heterosexual

1 0 1 0

28

Ground truth

5-Factor ModelOpenness

Conscientiousness

Extraversion

Agreeableness

Stability

29

Ground truth

Satisfaction with Life (SWL)

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 19: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

19

Prediction Accuracy

20

21

22

Language Use in Tweets

23

Social media opens up a new window of what humans actually feel and think

24

YOU ARE WHAT YOU LIKE

Social Psychology

Empirical Modeling

Kosinski Michal David Stillwell and Thore Graepel Private traits and attributes are predictable from digital records of human behavior Proceedings of the National Academy of Sciences 11015 (2013) 5802-5805

25

Personality Prediction

Personality traitsGender age relationship status friends

Sexual orientation ethnicity religion political inclination

Addictive substances (alcohol drugs cigarette) parental separation

IQ 5-Factor model satisfaction with Life

26

Data Collection

9939220 Likes (55814 unique ones) from 58466 Facebook volunteers

Sports

Music

Books

Restaurants

Popular websites

27

Ground truth

Political Inclination

Sexual Orientation

Democrat Republican

Democratic GOP (Grand Old Party)

Democratic Party Republican Party

Homosexual Heterosexual

1 0 1 0

28

Ground truth

5-Factor ModelOpenness

Conscientiousness

Extraversion

Agreeableness

Stability

29

Ground truth

Satisfaction with Life (SWL)

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 20: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

20

21

22

Language Use in Tweets

23

Social media opens up a new window of what humans actually feel and think

24

YOU ARE WHAT YOU LIKE

Social Psychology

Empirical Modeling

Kosinski Michal David Stillwell and Thore Graepel Private traits and attributes are predictable from digital records of human behavior Proceedings of the National Academy of Sciences 11015 (2013) 5802-5805

25

Personality Prediction

Personality traitsGender age relationship status friends

Sexual orientation ethnicity religion political inclination

Addictive substances (alcohol drugs cigarette) parental separation

IQ 5-Factor model satisfaction with Life

26

Data Collection

9939220 Likes (55814 unique ones) from 58466 Facebook volunteers

Sports

Music

Books

Restaurants

Popular websites

27

Ground truth

Political Inclination

Sexual Orientation

Democrat Republican

Democratic GOP (Grand Old Party)

Democratic Party Republican Party

Homosexual Heterosexual

1 0 1 0

28

Ground truth

5-Factor ModelOpenness

Conscientiousness

Extraversion

Agreeableness

Stability

29

Ground truth

Satisfaction with Life (SWL)

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 21: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

21

22

Language Use in Tweets

23

Social media opens up a new window of what humans actually feel and think

24

YOU ARE WHAT YOU LIKE

Social Psychology

Empirical Modeling

Kosinski Michal David Stillwell and Thore Graepel Private traits and attributes are predictable from digital records of human behavior Proceedings of the National Academy of Sciences 11015 (2013) 5802-5805

25

Personality Prediction

Personality traitsGender age relationship status friends

Sexual orientation ethnicity religion political inclination

Addictive substances (alcohol drugs cigarette) parental separation

IQ 5-Factor model satisfaction with Life

26

Data Collection

9939220 Likes (55814 unique ones) from 58466 Facebook volunteers

Sports

Music

Books

Restaurants

Popular websites

27

Ground truth

Political Inclination

Sexual Orientation

Democrat Republican

Democratic GOP (Grand Old Party)

Democratic Party Republican Party

Homosexual Heterosexual

1 0 1 0

28

Ground truth

5-Factor ModelOpenness

Conscientiousness

Extraversion

Agreeableness

Stability

29

Ground truth

Satisfaction with Life (SWL)

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 22: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

22

Language Use in Tweets

23

Social media opens up a new window of what humans actually feel and think

24

YOU ARE WHAT YOU LIKE

Social Psychology

Empirical Modeling

Kosinski Michal David Stillwell and Thore Graepel Private traits and attributes are predictable from digital records of human behavior Proceedings of the National Academy of Sciences 11015 (2013) 5802-5805

25

Personality Prediction

Personality traitsGender age relationship status friends

Sexual orientation ethnicity religion political inclination

Addictive substances (alcohol drugs cigarette) parental separation

IQ 5-Factor model satisfaction with Life

26

Data Collection

9939220 Likes (55814 unique ones) from 58466 Facebook volunteers

Sports

Music

Books

Restaurants

Popular websites

27

Ground truth

Political Inclination

Sexual Orientation

Democrat Republican

Democratic GOP (Grand Old Party)

Democratic Party Republican Party

Homosexual Heterosexual

1 0 1 0

28

Ground truth

5-Factor ModelOpenness

Conscientiousness

Extraversion

Agreeableness

Stability

29

Ground truth

Satisfaction with Life (SWL)

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 23: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

23

Social media opens up a new window of what humans actually feel and think

24

YOU ARE WHAT YOU LIKE

Social Psychology

Empirical Modeling

Kosinski Michal David Stillwell and Thore Graepel Private traits and attributes are predictable from digital records of human behavior Proceedings of the National Academy of Sciences 11015 (2013) 5802-5805

25

Personality Prediction

Personality traitsGender age relationship status friends

Sexual orientation ethnicity religion political inclination

Addictive substances (alcohol drugs cigarette) parental separation

IQ 5-Factor model satisfaction with Life

26

Data Collection

9939220 Likes (55814 unique ones) from 58466 Facebook volunteers

Sports

Music

Books

Restaurants

Popular websites

27

Ground truth

Political Inclination

Sexual Orientation

Democrat Republican

Democratic GOP (Grand Old Party)

Democratic Party Republican Party

Homosexual Heterosexual

1 0 1 0

28

Ground truth

5-Factor ModelOpenness

Conscientiousness

Extraversion

Agreeableness

Stability

29

Ground truth

Satisfaction with Life (SWL)

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 24: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

24

YOU ARE WHAT YOU LIKE

Social Psychology

Empirical Modeling

Kosinski Michal David Stillwell and Thore Graepel Private traits and attributes are predictable from digital records of human behavior Proceedings of the National Academy of Sciences 11015 (2013) 5802-5805

25

Personality Prediction

Personality traitsGender age relationship status friends

Sexual orientation ethnicity religion political inclination

Addictive substances (alcohol drugs cigarette) parental separation

IQ 5-Factor model satisfaction with Life

26

Data Collection

9939220 Likes (55814 unique ones) from 58466 Facebook volunteers

Sports

Music

Books

Restaurants

Popular websites

27

Ground truth

Political Inclination

Sexual Orientation

Democrat Republican

Democratic GOP (Grand Old Party)

Democratic Party Republican Party

Homosexual Heterosexual

1 0 1 0

28

Ground truth

5-Factor ModelOpenness

Conscientiousness

Extraversion

Agreeableness

Stability

29

Ground truth

Satisfaction with Life (SWL)

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 25: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

25

Personality Prediction

Personality traitsGender age relationship status friends

Sexual orientation ethnicity religion political inclination

Addictive substances (alcohol drugs cigarette) parental separation

IQ 5-Factor model satisfaction with Life

26

Data Collection

9939220 Likes (55814 unique ones) from 58466 Facebook volunteers

Sports

Music

Books

Restaurants

Popular websites

27

Ground truth

Political Inclination

Sexual Orientation

Democrat Republican

Democratic GOP (Grand Old Party)

Democratic Party Republican Party

Homosexual Heterosexual

1 0 1 0

28

Ground truth

5-Factor ModelOpenness

Conscientiousness

Extraversion

Agreeableness

Stability

29

Ground truth

Satisfaction with Life (SWL)

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 26: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

26

Data Collection

9939220 Likes (55814 unique ones) from 58466 Facebook volunteers

Sports

Music

Books

Restaurants

Popular websites

27

Ground truth

Political Inclination

Sexual Orientation

Democrat Republican

Democratic GOP (Grand Old Party)

Democratic Party Republican Party

Homosexual Heterosexual

1 0 1 0

28

Ground truth

5-Factor ModelOpenness

Conscientiousness

Extraversion

Agreeableness

Stability

29

Ground truth

Satisfaction with Life (SWL)

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 27: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

27

Ground truth

Political Inclination

Sexual Orientation

Democrat Republican

Democratic GOP (Grand Old Party)

Democratic Party Republican Party

Homosexual Heterosexual

1 0 1 0

28

Ground truth

5-Factor ModelOpenness

Conscientiousness

Extraversion

Agreeableness

Stability

29

Ground truth

Satisfaction with Life (SWL)

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 28: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

28

Ground truth

5-Factor ModelOpenness

Conscientiousness

Extraversion

Agreeableness

Stability

29

Ground truth

Satisfaction with Life (SWL)

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 29: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

29

Ground truth

Satisfaction with Life (SWL)

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 30: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

30

Methodology

User-Like matrix dimension reduction Singular Value Decomposition (SVD)

Prediction models Logistic Regression amp Linear Regression

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 31: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11 Author Michal Kosinski David Stillwell and Thore Graepel

Prediction ResultsSolid Pearson corr coef between pred amp actual valuesTransparent baseline acc of the questionnaire in terms of test-retest reliability

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 32: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

32

Discriminative Likes (1)

因版權疑慮移除相關資料

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 33: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

33

Discriminative Likes (2)

因版權疑慮移除相關資料

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 34: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

34

Discriminative Likes (3)

因版權疑慮移除相關資料

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 35: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

35

Likes are Culture-Dependent (1)

卡提諾正妹抱報 Catworld小舖

Garena《英雄聯盟 LOL》 QUEEN FASHION SHOP

遊戲大亂鬥 范范范瑋琪

Garena-TW 撿便宜特賣會

好色龍 衣芙日系

放棄治療 王大陸

這樣變型男 就愛網拍特賣會

Taipei Assassins (台北暗殺星) 86小舖商城

你為什麼要放棄治療呢 HH先生

Toyz LOVFEE

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 36: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

36

Likes are Culture-Dependent (2)

已婚

嬰兒與母親懷孕生產情報站 學生愛打工

未婚

味全 MyWei Duncan

Estee Lauder Taiwan 雅詩蘭黛 林俊傑 JJ Lin

光泉HOT鮮奶 Cherng

舒潔溫柔心感動 Byebyechuchu

綠巨人 Dcard

AVON Taiwan 雅芳粉絲團 田馥甄 Hebe

人人玩遊戲 彭于晏 Eddie Peng

Creative Baby - 台灣 Dorothy

阿默典藏蛋糕 韋禮安Weibird

Can we have real privacy on social mediaUnprecedented opportunity to observe individuals in a

society

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 37: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

別讓大數據變玄學

37

2016-03-08 0254經濟日報陳昇瑋

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 38: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

陳昇瑋 從資料科學到人工智慧

Big data vs Machine learning vs AI

Big data 3Vs

Machine learning ldquoA field of study that gives computers the ability to

learn without being explicitly programmed

Artificial intelligence

Turning test

38

真人

機器

Room 1Room 2

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 39: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

陳昇瑋 從資料科學到人工智慧 39

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 40: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

陳昇瑋 從資料科學到人工智慧 40

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 41: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

陳昇瑋 從資料科學到人工智慧 41

本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 42: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

陳昇瑋 從資料科學到人工智慧 42

如同精靈寶可夢需要有訓練師才能發揮能力擁有大數據

後我們也需要很多很多的機器學習專家(人稱「AI 訓練

師」)才能讓我們手中的大數據真正發揮價值

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 43: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

擁抱資料但別錯過 AI

陳昇瑋

台灣資料科學協會中央研究院資訊科學研究所

43

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 44: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

陳昇瑋 從資料科學到人工智慧

44

版權聲明

序 頁 作品 版權標章 作者來源

1 1Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

2 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo July 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

3 2

plosorg ldquoDynamics of Person-to-Person Interactions from Distributed RFID Sensor Networksrdquo uly 15 2010AuthorCiro Cattuto Wouter Van den Broeck Alain Barrat Vittoria Colizza Jean-Franccedilois Pinton Alessandro Vespignanihttpjournalsplosorgplosonearticleid=101371journalpone00115962017117 visited 依著作權法第465265條主張合理使用

4 3hbrorg ldquoThe New Science of Building Great Teamsrdquo by Alex Sandy Pentland FROM THE APRIL 2012 ISSUE 428httpshbrorg201204the-new-science-of-building-great-teams2017117 visited 依著作權法第465265條主張合理使用

556

7

Max pixel httpmaxpixelfreegreatpicturecomThumb-Hand-Viewing-Magnifying-Glass-Magnification-642672017117 visited

6 5

Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 45: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

陳昇瑋 從資料科學到人工智慧

45

版權聲明

序 頁 作品 版權標章 作者來源

7 5Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

8 8

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

9 9

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

10 10

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

11 11

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

12 12

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 46: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

陳昇瑋 從資料科學到人工智慧

46

版權聲明

序 頁 作品 版權標章 作者來源

13 13

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

14 14

plosorg ldquoPersonality Gender and Age in the Language of Social Media The Open-Vocabulary Approachrdquo September 25 2013AuthorH Andrew Schwartz Johannes C Eichstaedt Margaret L Kern Lukasz Dziurzynski Stephanie M Ramones Megha Agrawal AchalShah Michal Kosinski David Stillwell Martin E P Seligman Lyle H Ungarhttpjournalsplosorgplosonearticleid=101371journalpone0073791 2017117 visited 依著作權法第465265條主張合理使用

15 15Dreamstime Jasna01httpswwwdreamstimecomcomputer-room-stock-image-imagefree80551112017117 visited 依著作權法第465265條主張合理使用

16 16Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

17 18publicdomainpicturesnet Rf Vectorscomhttpwwwpublicdomainpicturesnetview-imagephpimage=581962017117 visited

18 18Wikimedia Commons AngelushttpscommonswikimediaorgwikiFileStub_doctorssvg2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 47: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

陳昇瑋 從資料科學到人工智慧

47

版權聲明

序 頁 作品 版權標章 作者來源

19 19

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

20202

1

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

21 22

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

22 22

The Authentic Happiness website ldquoPositive Psychology Theoryrdquohttpswwwauthentichappinesssasupennedulearn2017117 visited依著作權法第465265條主張合理使用

23 23

NCBI ldquoPsychological Language on Twitter Predicts County-Level Heart Disease Mortalityrdquo2015 Jan 20AuthorJohannes C Eichstaedt1 Hansen Andrew Schwartz12 Margaret L Kern13 Gregory Park1 Darwin R Labarthe4 Raina M Merchant5 Sneha Jha2 Megha Agrawal2 Lukasz A Dziurzynski1 Maarten Sap1 Christopher Weeg1 Emily E Larson1 Lyle H Ungar12 and Martin E P SeligmanhttpswwwncbinlmnihgovpmcarticlesPMC44335452017117 visited 依著作權法第465265條主張合理使用

24 24Flickr Rhett and LinkrhettandlinkhttpsflickrpJ8G99n2017117 visited

版權聲明

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 48: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

陳昇瑋 從資料科學到人工智慧

48

版權聲明

序 頁 作品 版權標章 作者來源

25 25Flickr Global PanoramatheglobalpanoramahttpsflickrpqbqQQk2017117 visited

26 26MaxpixelhttpmaxpixelfreegreatpicturecomHuman-Funny-Isolated-Holding-Female-Eye-Glass-156992017117 visited

27 26MaxpixelhttpmaxpixelfreegreatpicturecomFigures-Workers-Toys-Hauling-Dolls-Scoop-Broom-780022017117 visited

28 28

International Personality Item PoolhttpipiporiorgNew_IPIP-50-item-scalehtm2017117 visited依著作權法第465265條主張合理使用

29 29

midssorgldquoThe Satisfaction with Life Scale (SWL)rdquo Pavot W amp Diener E Files Scalehttpwwwmidssorgcontentsatisfaction-life-scale-swl2017117 visited依著作權法第465265條主張合理使用

30 30

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 49: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

陳昇瑋 從資料科學到人工智慧

49

版權聲明

序 頁 作品 版權標章 作者來源

31 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

32 31

NCBI ldquoPrivate traits and attributes are predictable from digital records of human behaviorrdquo2013 Mar 11Author Michal Kosinski David Stillwell and Thore GraepelhttpswwwncbinlmnihgovpmcarticlesPMC36253242017117 visited 依著作權法第465265條主張合理使用

33 35 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

34 36 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

35 37

Facebook Dan Arielyhttpswwwfacebookcomdanarielyposts9043835958682017117 visited依著作權法第465265條主張合理使用

36 37

經濟日報 ldquo別讓大數據變玄學rdquo 2016-03-08 0254經濟日報陳昇瑋httpmoneyudncommoneystory562915481412017117 visited依著作權法第465265條主張合理使用

版權聲明

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 50: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

陳昇瑋 從資料科學到人工智慧

50

版權聲明

序 頁 作品 版權標章 作者來源

37 38publicdomainpicturesnet Dawn Hudsonhttpwwwpublicdomainpicturesnetview-imagephpimage=134740amppicture=computer-guy2017117 visited

38 38Flickr Steve WOCinTech ChatwocintechchatcomhttpsflickrpER8emC2017117 visited

39 38Wikimedia Commons PaulVernon1974httpscommonswikimediaorgwikiFileAcornArchimedes-Wikijpg2017117 visited

40 39 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

41 40 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

42 41 本作品由陳昇瑋授權使用本中心無再授權他人使用之權利如需使用請另行向權利人取得授權

版權聲明

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明

Page 51: Social Psychology YOU ARE WHAT YOU LIKEget.aca.ntu.edu.tw/getcdb/retrieve/448982/17101L03_201808.pdf · Kosinski, Michal, David Stillwell, and Thore Graepel. "Private traits and attributes

陳昇瑋 從資料科學到人工智慧

51

版權聲明

序 頁 作品 版權標章 作者來源

43 42Flickr Saad AkhtarSaadAkhtarhttpsflickrp2cj282017117 visited

44 43

台灣資料科學愛好者年會httpdatascitw20152017117 visited依著作權法第465265條主張合理使用

版權聲明