ecir 2013 keynote - time for events

71
time for events telling the world’s stories from social media Mor Naaman Rutgers SC&I & Mahaya, Inc. @informor

Upload: mor

Post on 28-Nov-2014

2.621 views

Category:

Technology


0 download

DESCRIPTION

 

TRANSCRIPT

Page 1: ECIR 2013 Keynote - Time for Events

time for events telling the world’s stories from social media

Mor Naaman Rutgers SC&I & Mahaya, Inc.

@informor

Page 2: ECIR 2013 Keynote - Time for Events
Page 3: ECIR 2013 Keynote - Time for Events
Page 4: ECIR 2013 Keynote - Time for Events

enter: social media

Page 5: ECIR 2013 Keynote - Time for Events

(JCDL 2007)

Page 6: ECIR 2013 Keynote - Time for Events

(JCDL 2007)

Page 7: ECIR 2013 Keynote - Time for Events

yes.

(SIGIR 2007)

Page 8: ECIR 2013 Keynote - Time for Events

organize the world’s memories

Page 9: ECIR 2013 Keynote - Time for Events

people, together

Page 10: ECIR 2013 Keynote - Time for Events

BYOBW

Page 11: ECIR 2013 Keynote - Time for Events

outside lands festival

Page 12: ECIR 2013 Keynote - Time for Events
Page 13: ECIR 2013 Keynote - Time for Events
Page 14: ECIR 2013 Keynote - Time for Events
Page 15: ECIR 2013 Keynote - Time for Events

organize the world’s memories

Page 16: ECIR 2013 Keynote - Time for Events

objectives d

ete

ct

ide

nti

fy

org

an

ize

Page 17: ECIR 2013 Keynote - Time for Events

objectives d

ete

ct

ICWSM 2011a JASIST 2011 WebDB 2009 SIGIR 2007

Page 18: ECIR 2013 Keynote - Time for Events

objectives

ide

nti

fy

WSDM 2012 ICWSM 2011b WSDM 2010

Page 19: ECIR 2013 Keynote - Time for Events

objectives

org

an

ize

ICMR 2012 CHI 2012

CSCW 2012 MTAP 2012 VAST 2010

WWW 2009

Page 20: ECIR 2013 Keynote - Time for Events

today d

ete

ct

org

an

ize

Vox!Multiplayer

multi-site id

en

tify

Page 21: ECIR 2013 Keynote - Time for Events

Vox Civitas

over

view

Multiplayer

Multi-site content E

Page 22: ECIR 2013 Keynote - Time for Events

[with Hila Becker, Luis Gravano]

goal effectively retrieve social media content for known events from multiple services

E

Page 23: ECIR 2013 Keynote - Time for Events

E

Page 24: ECIR 2013 Keynote - Time for Events

challenges event descriptor not well-formed brief textual descriptors noise formats/conventions/metadata differ

E

Page 25: ECIR 2013 Keynote - Time for Events

approach two-step query formulation

precision-based recall-based

validate queries based on known/extracted event model

E

Page 26: ECIR 2013 Keynote - Time for Events

step 1 term extraction from event descriptors generates “high precision” queries e. g. “andrew bird, opening gala, celebrate brooklyn, prospect park”

E E

Page 27: ECIR 2013 Keynote - Time for Events

step 2 use “high precision” corpus to generate more general queries to improve recall e. g. “andrew bird concert”, “state farm insurance”

E E

Page 28: ECIR 2013 Keynote - Time for Events

recall-oriented queries Benefits: - Works cross-site - Works with short content Challenges: - Introduces noise - Potentially large set of queries

E E

Page 29: ECIR 2013 Keynote - Time for Events

post-filtering use known event model (topics, time, location) use queries with a result set that matches known model

E E

Page 30: ECIR 2013 Keynote - Time for Events

for example...

E E

0"20"40"60"80"

100"120"

6/7/11" 6/8/11" 6/9/11" 6/10/11" 6/11/11" 6/12/11" 6/13/11"

[andrew"bird"concert]" [state"farm"insurance]"

Page 31: ECIR 2013 Keynote - Time for Events

5" 5"

4" 4"

39" 36" 34" 34"

9" 8" 8" 7"

0"0.1"0.2"0.3"0.4"0.5"0.6"0.7"0.8"0.9"1"

1.1"

0" 5" 10" 15" 20" 25"

NDC

G%

Number%of%Documents%k%

Precision"

Twi7er8MS"

YouTube8MS"

evaluation query generation relevance of retrieved documents

E

Page 32: ECIR 2013 Keynote - Time for Events

takeaways can aggregate content fragmented across platforms improve recall, not rely on site-specific features

E

Page 33: ECIR 2013 Keynote - Time for Events

Vox Civitas

over

view

Multiplayer

Multi-site content E (WSDM 2012)

Page 34: ECIR 2013 Keynote - Time for Events
Page 35: ECIR 2013 Keynote - Time for Events

[with postdoctoral fellow Nick Diakopoulos]

research questions can Twitter content around broadcast news events inform journalistic inquiry? what insights and analyses can we enable through visual analytic tools?

Page 36: ECIR 2013 Keynote - Time for Events

direct attention to relevant information

automatic content analysis for filtering

– relevance

– uniqueness / novelty

– sentiment

– keyword extraction

supporting analysis

Page 37: ECIR 2013 Keynote - Time for Events
Page 38: ECIR 2013 Keynote - Time for Events

how to evaluate? directly evaluate the output of the algorithms (quantitative)

deep, extensive evaluation of users’ interaction with the system (qualitative)  

read more: Olsen (UIST ’07) Naaman (MTAP ’12)

Page 39: ECIR 2013 Keynote - Time for Events

Vox evaluation goals •  How effective for generating story ideas?

•  What kind of insights/analysis are supported?

•  Shortcomings and how features are used?

Page 40: ECIR 2013 Keynote - Time for Events

takeaways can extract reliable event structure from social media

Page 41: ECIR 2013 Keynote - Time for Events

Vox Civitas

over

view

Multiplayer

Multi-site content E

(VAST 2010)

Page 42: ECIR 2013 Keynote - Time for Events

what the hell?

[with: Lyndon Kennedy, Dan Ellis, Kai Su]

Page 43: ECIR 2013 Keynote - Time for Events
Page 44: ECIR 2013 Keynote - Time for Events
Page 45: ECIR 2013 Keynote - Time for Events
Page 46: ECIR 2013 Keynote - Time for Events
Page 47: ECIR 2013 Keynote - Time for Events

supporting analysis extract the signal from people’s attention: find overlapping moments compute and rank scenes extract scene descriptors

Page 48: ECIR 2013 Keynote - Time for Events

audio fingerprinting

Wang et al. (ISMIR ’03)

Page 49: ECIR 2013 Keynote - Time for Events

two clips, aligned

0:00

0:00 0:18

2:32

3:32

Page 50: ECIR 2013 Keynote - Time for Events

a story of n clips

time

Page 51: ECIR 2013 Keynote - Time for Events

from clips to scenes

time Happy Birthday, Birthday

Higher Ground Encore

Page 52: ECIR 2013 Keynote - Time for Events
Page 53: ECIR 2013 Keynote - Time for Events

evaluation quantitative: evaluated matching, scene extraction… qualitative: evaluated deployment scenario/task

Page 54: ECIR 2013 Keynote - Time for Events

takeaways can create an event presentation that gets better them more content is added

Page 55: ECIR 2013 Keynote - Time for Events

Vox Civitas

over

view

Multiplayer

Multi-site content E

(NM&S 2012, ICMR 2012, MTAP 2012, WWW 2009)

Page 56: ECIR 2013 Keynote - Time for Events
Page 57: ECIR 2013 Keynote - Time for Events

towards better models of large-scale human attention

Page 58: ECIR 2013 Keynote - Time for Events

printing press

Page 59: ECIR 2013 Keynote - Time for Events

è knowledge archive

Page 60: ECIR 2013 Keynote - Time for Events

digital documents

Page 61: ECIR 2013 Keynote - Time for Events

èdigital archive

Page 62: ECIR 2013 Keynote - Time for Events

the web

Page 63: ECIR 2013 Keynote - Time for Events

ènetworked archive

Page 64: ECIR 2013 Keynote - Time for Events

social media

Page 65: ECIR 2013 Keynote - Time for Events

èexperience archive

Page 66: ECIR 2013 Keynote - Time for Events

new methods?

Page 67: ECIR 2013 Keynote - Time for Events

search by subject code?

Page 68: ECIR 2013 Keynote - Time for Events

explore. new information seeking tasks (and models) new applications for social media content

Page 69: ECIR 2013 Keynote - Time for Events

explore.

beyond real-time personal and social

Page 70: ECIR 2013 Keynote - Time for Events

[email protected] @informor

http://mornaaman.com

questions?

Page 71: ECIR 2013 Keynote - Time for Events

Luis Gravano Hila Becker Nick Diakopoulos Kai Su Dan Ellis Munmun de Choudhury Tarikh Korula …

thanks