Elamur Research⌕
☰
home / papers / 2011 / Unsupervised Discovery of Discourse Relations for Eliminating Intra-sentence Polarity Ambiguities

Unsupervised Discovery of Discourse Relations for Eliminating Intra-sentence Polarity Ambiguities

Summary

Sentences that mix positive and negative opinions are hard for sentence-level polarity classifiers. A sentence can carry more negative words than positive ones and still be positive overall. This paper uses the discourse structure inside a sentence to decide which clause controls the overall sentiment. It targets Chinese news text from the NTCIR MOAT corpora.

The authors derive a coarse scheme from Rhetorical Structure Theory with five relations: Contrast, Condition, Continuation, Cause and Purpose. Each relation comes with rules about how the polarities of its segments interact. For example, the nucleus decides the polarity for most relations.

Starting from a handful of cue-phrase patterns, they harvest many clause pairs from raw text. These pairs are encoded as semantic sequential representations (SSRs) built from part-of-speech, semantic and sentiment tags. An unsupervised learner then generates, weights and filters common SSRs, so that relations can be recognized even when no cue phrase is present. The SSRs also serve as SVM features.

Adding the recovered discourse information gives a statistically significant gain over a strong baseline. Because no annotated discourse treebank is required, the approach should transfer to other languages.

Key points

  • About 43% of opinionated sentences in the NTCIR-6/7 MOAT Chinese corpora contain mixed polarity. Ambiguous sentences account for 49% of the errors made by the baseline polarity classifier (BPC), which is the Xu and Kit (2010) system from NTCIR-8 MOAT.
  • The scheme was built by annotating 500 ambiguous sentences. Thirteen RST relations covered roughly 70% of the useful discourse links, and these were merged into 5 coarse relations.
  • Clauses are split at commas and semicolons. Each harvested segment pair must contain at least one sentiment word per segment.
  • Weighting and filtering of mined SSRs (cSSR) raise precision from 0.6182 to 0.8864 and F-score from 0.5537 to 0.6293. They also shrink the SSR set from over 1 million to about 0.12 million, at a small cost in recall.
  • SVM+SSRs significantly outperform BPC in polarity classification (p<0.01). Most of the gains come from the Contrast, Continuation and Cause relations.

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: Linguistics & languages

Topics: Discourse analysis

Cited by

18 papers of our library cite it (of the 140673 whose bibliographies we read); 3 of them have a page here, 15 not yet.

Cite

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

@inproceedings{zhou2011unsupervised,
  title = {{Unsupervised Discovery of Discourse Relations for Eliminating Intra-sentence Polarity Ambiguities}},
  author = {Zhou, Lanjun and Li, Binyang and Gao, Wei and Wei, Zhongyu and Wong, Kam-Fai},
  booktitle = {Proceedings of the 2011 Conference on Empirical Methods in Natural Language Processing},
  year = {2011},
  url = {https://aclanthology.org/D11-1015/}
}

References

22 references, 8 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. N. Asher, F. Benamara, and Y.Y. Mathieu. 2008. Distilling opinion in discourse: A preliminary study. find
  2. Coling 2008: Companion volume: Posters and Demonstrations, pages 5--8. find
  3. S. Blair-Goldensohn, K.R. McKeown, and O.C. Rambow. 2007. Building and refining rhetorical-semantic relation models. In Proceedings of NAACL HLT, pages 428--435. find
  4. L. Carlson, D. Marcu, and M.E. Okurowski. 2001. Building a discourse-tagged corpus in the framework of rhetorical structure theory. In Proceedings of the Second SIGdial Workshop on Discourse and Dialogue-Volume 16, pages 1--10. Association for Computational Linguistics. find our card
  5. W. Che, Z. Li, and T. Liu. 2010. Ltp: A chinese language technology platform. In Proceedings of the 23rd International Conference on Computational Linguistics: Demonstrations, pages 13--16. Association for Computational Linguistics. find
  6. D.A. Duverle and H. Prendinger. 2009. A novel discourse parser based on support vector machine classification. 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: Volume 2, pages 665--673… find our card
  7. C. Fellbaum. 1998. WordNet: An electronic lexical database. The MIT press. find
  8. M. Hu and B. Liu. 2004. Mining and summarizing customer reviews. In Proceedings of the tenth ACM SIGKDD international conference on Knowledge discovery and data mining, pages 168--177. ACM. find
  9. W.C. Mann and S.A. Thompson. 1988. Rhetorical structure theory: Toward a functional theory of text organization. Text-Interdisciplinary Journal for the Study of Discourse, 8(3):243--281. find our card
  10. D. Marcu and A. Echihabi. 2002. An unsupervised approach to recognizing discourse relations. In Proceedings of the 40th Annual Meeting on Association for Computational Linguistics, pages 368--375. Association for Computational Linguistics. find our card
  11. E. Miltsakaki, R. Prasad, A. Joshi, and B. Webber. 2004. The penn discourse treebank. In Proceedings of the 4th International Conference on Language Resources and Evaluation. Citeseer. find our card
  12. B. Pang, L. Lee, and S. Vaithyanathan. 2002. Thumbs up?: sentiment classification using machine learning techniques. In Proceedings of the ACL-02 conference on Empirical methods in natural language processing-Volume 10, pages 79--86. Association for Computational Linguistics. find
  13. L. Polanyi and A. Zaenen. 2006. Contextual valence shifters. Computing attitude and affect in text: Theory and applications, pages 1--10. find
  14. E. Riloff, J. Wiebe, and T. Wilson. 2003. Learning subjective nouns using extraction pattern bootstrapping. In Proceedings of the seventh conference on Natural language learning at HLT-NAACL 2003-Volume 4, pages 25--32. Association for Computational Linguistics. find our card
  15. K. Sadamitsu, S. Sekine, and M. Yamamoto. 2008. Sentiment analysis based on probabilistic models using inter-sentence information. find
  16. B. Snyder and R. Barzilay. 2007. Multiple aspect ranking using the good grief algorithm. In Proceedings of NAACL HLT, pages 300--307. find
  17. S. Somasundaran, J. Wiebe, and J. Ruppenhofer. 2008. Discourse level opinion interpretation. In Proceedings of the 22nd International Conference on Computational Linguistics, pages 801--808. Association for Computational Linguistics. find
  18. S. Somasundaran, G. Namata, J. Wiebe, and L. Getoor. 2009. Supervised and unsupervised methods in employing discourse relations for improving opinion polarity classification. In Proceedings of the 2009 Conference on Empirical Methods in Natural Language Processing, pages 170--179. Association for Co… find our card
  19. R. Soricut and D. Marcu. 2003. Sentence level discourse parsing using syntactic and lexical information. In Proceedings of the 2003 Conference of the North American Chapter of the Association for Computational Linguistics on Human Language Technology, pages 149--156. Association for Computational Li… find our card
  20. T. Wilson, J. Wiebe, and P. Hoffmann. 2009. Recognizing Contextual Polarity: an exploration of features for phrase-level sentiment analysis. Computational Linguistics, 35(3):399--433. find
  21. Y.Q. Xia, R.F. Xu, K.F. Wong, and F. Zheng. 2007. The unified collocation framework for opinion mining. In International Conference on Machine Learning and Cybernetics, volume 2, pages 844--850. IEEE. find
  22. R. Xu and C. Kit. 2010. Incorporating feature-based and similarity-based opinion mining--ctl in ntcir-8 moat. In Proceedings of the 8th NTCIR Workshop, pages 276--281. find

↑ back to top

This page in Markdown →