sentiment analysis (thanks to matt baker). introduction what how conclusion laptop purchase how will...

Post on 19-Jan-2016

221 Views

Category:

Documents

4 Downloads

Preview:

Click to see full reader

TRANSCRIPT

Sentiment Analysis

(thanks to Matt Baker)

IntroductionWhatHowConclusion

Laptop Purchase

How will you decide?

IntroductionWhatHowConclusion

Survey Says

81% internet users

• online product research 1+ times

20% internet users

• online product research daily

73-87% consumers of restaurant, hotel, service reviews

• reviews significantly impact purchasing decisionscomScore/the Kelsey group, “Online consumer-generated reviews have significant impact on offline purchase behavior,” Press Release, http://www.comscore.com/press/release.asp?press=1928, November 2007.Quoted in Pang and Lee, 2008, “Opinion Mining and Sentiment Analysis”

IntroductionWhatHowConclusion

Survey Says

20-99% consumers

• willing to pay more for 5- than 4-star-rated product

32% consumers

• rated online product, service, person

30% consumers

• posted online comment or reviewJ. A. Horrigan, “Online shopping,” Pew Internet & American Life Project Report, 2008.Quoted in Pang and Lee, 2008, “Opinion Mining and Sentiment Analysis”

Introduction WhatHowConclusion

Definition

Many scholars view sentiment analysis as “the computational treatment of opinion,

sentiment, and subjectivity in text”*

How do each of these terms differ?

*Pang 2008, pg. 8

Introduction WhatHowConclusion

Review of Literature

• Uses machine learning to assess polarity (positive, negative, neutral) in movie reviews– Pang, Lee, & Vaithyanathan (2002)

• Evaluates sentiment based on parts of speech (adjectives and adverbs)– Turney (2002)

Introduction WhatHowConclusion

Review of Literature

• Separates objective from subjective statements and assess polarity of opinion sentences– Yu & Hatzivassiloglou (2003)

• Identifies valence shifters in text that can give information regarding the writer’s sentiment– Polanyi & Zaenen (2004)

Introduction WhatHowConclusion

Review of Literature

• Expands sentiment analysis to include rankings on a scale– Pang & Lee (2005)

• Selects features from text and performs sentiment analysis on a feature level– Durant & Smith (2007)

IntroductionWhat HowConclusion

Twitter

How would you extract sentiment from Tweets?

IntroductionWhat HowConclusion

Considerations

• Parts of speech• Objective statements• Subjective statements• Binary classification• Ranking• Features• Overall sentiment• Domain• Word position

IntroductionWhat HowConclusion

Considerations• Valence shifters*

– Words– Negation– Intensifiers– Modal operators– Irony

• Pronoun resolution• Topic relevance• Unigrams, bigrams, etc.• Syntax• Strength of polarity

*Polanyi & Zaenen (2004)

IntroductionWhat HowConclusion

Twitter Literature

• Sentiment word frequencies*• Emoticons ***• Unigrams***• Bigrams***• Parts of Speech***

*O’Conner et al. 2010***Go 2009

IntroductionWhat HowConclusion

Sample reviews (negative polarity)

• a peculiar misfire that even tunney can't save . • watching queen of the damned is like reading a

research paper , with special effects tossed in . • i can't remember the last time i saw worse stunt

editing or cheaper action movie production values than in extreme ops .

• too much of nemesis has a tired , talky feel . • i felt trapped and with no obvious escape for the

entire 100 minutes . • a baffling mixed platter of gritty realism and

magic realism with a hard-to-swallow premise . • an affable but undernourished romantic comedy

that fails to match the freshness of the actress-producer and writer's previous collaboration , miss congeniality

IntroductionWhat HowConclusion

Sample reviews (positive polarity)

• emerges as something rare , an issue movie that's so honest and keenly observed that it doesn't feel like one .

• the film provides some great insight into the neurotic mindset of all comics -- even those who have reached the absolute top of the game .

• offers that rare combination of entertainment and education . • perhaps no picture ever made has more literally showed that

the road to hell is paved with good intentions . • offers a breath of the fresh air of true sophistication . • a thoughtful , provocative , insistently humanizing film . • not for everyone , but for those with whom it will connect ,

it's a nice departure from standard moviegoing fare .• is it a total success ? no . is it something any true film

addict will want to check out ? you bet . • engrossing and affecting , if ultimately not quite satisfying

.

IntroductionWhat HowConclusion

Lexical characterizations

IntroductionWhat HowConclusion

Tools: SentiWordNet

http://sentiwordnet.isti.cnr.it/

IntroductionWhat HowConclusion

Tools: Opinon Lexicon

http://www.cs.uic.edu/~liub/FBS/sentiment-analysis.html#lexicon

IntroductionWhat HowConclusion

Online data / demos / tools

• Movie review data• NLTK

IntroductionWhat HowConclusion

Systems

• LingPipe: tutorial and system you can install

• Weka: blog and instructions • R: blog and pointer to code

IntroductionWhat HowConclusion

Competitions

Kaggle.com

IntroductionWhat HowConclusion

Methods: Linear Regression

IntroductionWhat HowConclusion

Example

Twitter with R*

*https://github.com/jeffreybreen/twitter-sentiment-analysis-tutorial-201107/blob/master/R/0_start.R*http://www.inside-r.org/howto/mining-twitter-airline-consumer-sentiment

IntroductionWhatHow Conclusion

Other applications

• Classifying speeches as for or against issues*• Discovering the political leanings of texts**• Sensing user annoyance by computers to

change interaction methods***• Monitoring violent and hateful

propaganda****• Tracking the world’s mood*****• Scanning emails

*Thomas, et al. 2006**Pang 2008***Liscombe et al. 2005****Abassi 2007

References• A. Abbasi, “Affect intensity analysis of dark web forums,” in Proceedings of Intelligence and Security Informatics (ISI),

pp. 282–288, 2007.• Bo, Pang, and Lillian Lee. "Opinion Mining and Sentiment Analysis." Foundations & Trends in Information Retrieval 2,

no. 1/2 (2008): 1-135.• Bo, Pang, Lillian Lee, and Shivakumar Vaithyanathan. “Thumbs up? Sentiment Classification using Machine Learning

Techniques. Proceedings of the Conference on Empirical Methods in Natural Language Processing. (2002).• Durant, Kathleen and Michael Smith. “Predicting the Political Sentiment of Web Log Posts using Supervised Machine

Learning Techniques Coupled with Feature Selection.” 2007.• Go, Alec. “Twitter Sentiment Analysis.” 2009.• J. Liscombe, G. Riccardi, and D. Hakkani-T¨ur, “Using context to improve emotion detection in spoken dialog systems,”

in Interspeech, pp. 1845–1848, 2005.• M. Thomas, B. Pang, and L. Lee, “Get out the vote: Determining support or opposition from congressional floor-debate

transcripts,” in Proceedings of the Conference on Empirical Methods in Natural Language Processing (EMNLP), pp. 327–335, 2006.

• O’Connor, Brandon, Ramnath Balasubramanyan, Bryan R. Routledge, and Noah A. Smith. “From Tweets to Polls: Linking Text Sentiment to Public Opinion Time Series.” 2010.

• Pang, Bo and Lillian Lee. “Seeing stars: Exploiting class relationships for sentiment categorization with respect to rating scales.” 2005.

• Polanyi, Livia and Annie Zaenen. “Contextual valence shifters.” Computing attitude and affect in text: Theory and applications. 2006.

• Turney, Peter. “Thumbs up or Thumbs Down? Semantic Orientation Applied to Unsupervised Classification of Reviews.” Proceedings of the 40th Annual Meeting of the Association for Computational Linguistics. (2002).

• Yu, Hong and Vasileios Hatzivassiloglou. “Towards answering opinion questions: separating facts from opinions and identifying the polarity of opinion sentences.” Proceedings of the 2003 conference on Empirical methods in natural language processing. 2003.

top related