Elamur Research⌕
☰
home / papers / 2010 / Recognizing Stances in Ideological On-Line Debates

Recognizing Stances in Ideological On-Line Debates

Summary

The paper asks whether opinion analysis can tell which side a writer takes in two-sided ideological debates on the web, such as abortion or gun rights. The authors distinguish two kinds of subjectivity. Sentiment covers positive or negative evaluations. Arguing covers claims about what is true or what ought to be done. Arguing had no existing lexical resource, so they derive one automatically from the arguing annotations in the MPQA corpus. They extract the leading n-grams that act as arguing "triggers."

Each opinion is then paired with what it is about, producing opinion-target features for a supervised stance classifier. Posts were collected from debate websites, and debate-specific sides were mapped by hand onto domain-level for/against stances. Several systems are compared: sentiment-only, arguing-only, a combination of both, a unigram baseline and a distribution-based baseline. The combined system clearly beats the distribution baseline and slightly outperforms unigrams. Arguing features alone match unigrams, while sentiment alone falls short. Feature inspection suggests the opinion-based features expose the underlying ideological reasoning better than bare words.

Key points

  • The corpus covers six domains collected from online debate sites, each with posts carrying user-provided stance tags:
    • Existence of God and Healthcare are used for development.
    • Gun Rights, Gay Rights, Abortion and Creationism are used for experiments.
  • Domain sizes range from 336 posts (Healthcare, 16 debates) to 1186 posts (Gay Rights, 15 debates). The other domains are:
    • Existence of God: 486 posts
    • Gun Rights: 566 posts
    • Abortion: 618 posts
    • Creationism: 729 posts
  • The arguing lexicon is built from MPQA corpus (Version 2) positive and negative arguing annotations. It contains n-grams up to trigrams, taken mainly from the start of annotated spans.
  • Sentiment clues come from the existing lexicon of Wilson et al. (2005).
  • Features take forms like "ap-rights" (positive arguing toward "rights") or "an-gay" (negative arguing toward "gay").
  • The Arg+Sent system performs best overall, slightly above the unigram system. Arguing-only is on par with unigrams, and sentiment-only is weaker than unigrams.

Interesting ideas

  • I1. Splitting annotations into trigger and remainder. The authors treat the start of each annotated span as the part that signals the opinion type and discard the rest. This makes span-level annotations usable as a general clue lexicon. The same approach could turn other span-annotated corpora into reusable lexicons for languages or categories that lack them.
  • I2. Pairing opinions with their targets. The polarity of an opinion, or the topic mentioned, is often ambiguous on its own. Combining the two into one feature resolves this, for example negativity toward government versus negativity toward insurance companies. This is a cheap trick for any task where meaning depends on what an evaluation is directed at.
  • I3. Mapping local labels onto shared global labels. Each debate has its own side labels ("Yes!!", "for", "It has not"), which the authors map manually onto shared domain-level stances. This lets many small, heterogeneously labelled sources be pooled into one dataset, which is useful when scraping forums or polls.
  • I4. Inspecting features to test interpretability. The authors list the top features per stance to show what each feature type captures, not just to report scores. The analysis shows that similar concepts receive the same polarity, while alternative concepts receive opposite polarity for the same side. Such qualitative checks help justify a richer feature set even when its accuracy gain is modest.
  • I5. Separating development and test domains. Some topics are reserved for designing the approach and others for evaluation. This reduces the risk of tuning to particular topics and is a good practice for cross-domain classification studies.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

Fields: Natural language processing

Cited by

123 papers of our library cite it (of the 140884 whose bibliographies we read); 21 of them have a page here, 102 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work.

@inproceedings{somasundaran2010recognizing,
  title = {{Recognizing Stances in Ideological On-Line Debates}},
  author = {Somasundaran, Swapna and Wiebe, Janyce},
  booktitle = {Proceedings of the NAACL HLT 2010 Workshop on Computational Approaches to Analysis and Generation of Emotion in Text},
  year = {2010},
  url = {https://aclanthology.org/W10-0214/}
}

References

24 references, 2 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Lada A. Adamic and Natalie Glance. 2005. The political blogosphere and the 2004 u.s. election: Divided they blog. In LinkKDD. find
  2. Rakesh Agrawal, Sridhar Rajagopalan, Ramakrishnan Srikant, and Yirong Xu. 2003. Mining newsgroups using networks arising from social behavior. In WWW. find
  3. Mohit Bansal, Claire Cardie, and Lillian Lee. 2008. The power of negative thinking: Exploiting label disagreement in the min-cut classification framework. In Proceedings of the 22nd International Conference on Computational Linguistics (COLING-2008). find
  4. Yejin Choi and Claire Cardie. 2009. Adapting a polarity lexicon using integer linear programming for domain-specific sentiment classification. In Proceedings of the 2009 Conference on Empirical Methods in Natural Language Processing, pages 590–598, Singapore, August. Association for Computational Li… find
  5. Miles Efron. 2004. Cultural orientation: Classifying subjective documents by cocitation analysis. In AAAI Fall Symposium on Style and Meaning in Language, Art, and Music. find
  6. Stephan Greene and Philip Resnik. 2009. More than words: Syntactic packaging and implicit sentiment. In Proceedings of Human Language Technologies: The 2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics, pages 503–511, Boulder, Colorado, June. Assoc… find
  7. Gregory Grefenstette, Yan Qu, James G. Shanahan, and David A. Evans. 2004. Coupling niche browsers and affect analysis for an opinion mining application. In Proceeding of RIAO-04, Avignon, FR. find
  8. Mark Hall, Eibe Frank, Geoffrey Holmes, Bernhard Pfahringer, Peter Reutemann, and Ian H. Witten. 2009. The weka data mining software: An update. In SIGKDD Explorations, Volume 11, Issue 1. find
  9. Soo-Min Kim and Eduard Hovy. 2007. Crystal: Analyzing predictive opinions on the web. In Proceedings of the 2007 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning (EMNLP-CoNLL), pages 1056–1064. find
  10. Michael Laver, Kenneth Benoit, and John Garry. 2003. Extracting policy positions from political texts using words as data. American Political Science Review, 97(2):311–331. find
  11. Wei-Hao Lin, Theresa Wilson, Janyce Wiebe, and Alexander Hauptmann. 2006. Which side are you on? Identifying perspectives at the document and sentence levels. In Proceedings of the 10th Conference on Computational Natural Language Learning (CoNLL-2006), pages 109–116, New York, New York. find
  12. Wei-Hao Lin. 2006. Identifying perspectives at the document and sentence levels using statistical models. In Proceedings of the Human Language Technology Conference of the NAACL, Companion Volume: Doctoral Consortium, pages 227–230, New York City, USA, June. Association for Computational Linguistics… find
  13. Robert Malouf and Tony Mullen. 2008. Taking sides: Graph-based user classification for informal online political discourse. Internet Research, 18(2). find
  14. Lanny W. Martin and Georg Vanberg. 2008. A robust transformation procedure for interpreting political text. Political Analysis, 16(1):93–100. find
  15. Tony Mullen and Robert Malouf. 2006. A preliminary investigation into sentiment analysis of informal political discourse. In AAAI 2006 Spring Symposium on Computational Approaches to Analysing Weblogs (AAAI-CAAW 2006). find
  16. Bo Pang and Lillian Lee. 2008. Opinion mining and sentiment analysis. Foundations and Trends in Information Retrieval, Vol. 2(1-2):pp. 1–135. find
  17. Swapna Somasundaran and Janyce Wiebe. 2009. Recognizing stances in online debates. In Proceedings of the Joint Conference of the 47th Annual Meeting of the ACL and the 4th International Joint Conference on Natural Language Processing of the AFNLP, pages 226–234, Suntec, Singapore, August. Associatio… find
  18. Swapna Somasundaran, Josef Ruppenhofer, and Janyce Wiebe. 2007a. Detecting arguing and sentiment in meetings. In SIGdial Workshop on Discourse and Dialogue, Antwerp, Belgium, September. find
  19. Swapna Somasundaran, Theresa Wilson, Janyce Wiebe, and Veselin Stoyanov. 2007b. Qa with attitude: Exploiting opinion type analysis for improving question answering in on-line discussions and the news. In International Conference on Weblogs and Social Media, Boulder, CO. find
  20. Swapna Somasundaran, Janyce Wiebe, and Josef Ruppenhofer. 2008. Discourse level opinion interpretation. In Proceedings of the 22nd International Conference on Computational Linguistics (Coling 2008), pages 801–808, Manchester, UK, August. find
  21. Matt Thomas, Bo Pang, and Lillian Lee. 2006. Get out the vote: Determining support or opposition from congressional floor-debate transcripts. In Proceedings of the 2006 Conference on Empirical Methods in Natural Language Processing, pages 327–335, Sydney, Australia, July. Association for Computation… find our card
  22. Theresa Wilson and Janyce Wiebe. 2005. Annotating attributions and private states. In Proceedings of ACL Workshop on Frontiers in Corpus Annotation II: Pie in the Sky. find
  23. Theresa Wilson, Janyce Wiebe, and Paul Hoffmann. 2005. Recognizing contextual polarity in phrase-level sentiment analysis. In hltemnlp2005, pages 347–354, Vancouver, Canada. find our card
  24. Theresa Wilson. 2007. Fine-grained Subjectivity and Sentiment Analysis: Recognizing the Intensity, Polarity, and Attitudes of private states. Ph.D. thesis, Intelligent Systems Program, University of Pittsburgh. find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/somasundaran-2010-recognizing-stances-ideological-line-debates/