Elamur Research
☰
home / papers / 2015 / Harnessing Context Incongruity for Sarcasm Detection

Harnessing Context Incongruity for Sarcasm Detection

Written for the Research library (claude-opus-5-5) from the paper text, in our own words: the license "CC BY-NC-SA 3.0 (ACL Anthology, before 2016)" does not allow redistributing the text itself. Original — aclanthology.org/P15-2124.pdf

Summary

Linguists have long argued that sarcasm comes from a mismatch between what is said and what the context leads us to expect. This mismatch is called context incongruity. Joshi, Sharma and Bhattacharyya turn this idea into features for a supervised sarcasm classifier. They separate two cases. In explicit incongruity, words of opposite sentiment polarity appear together. In implicit incongruity, a polar word conflicts with a phrase whose sentiment is only implied, such as an unpleasant situation. The explicit features come from earlier work on detecting thwarted expectations. The implicit features are sentiment-bearing verb and noun phrases, mined with an extended version of the Riloff et al. (2013) bootstrapping algorithm. The authors add these to unigram and pragmatic cues in an SVM. They test it on two tweet collections and a sample of the Internet Argument Corpus. The new features raise F-score over a unigram baseline for both tweets and forum posts, and the system beats two published systems by 10–20 F-score points. The paper bases sarcasm detection on linguistic theory and makes an early attempt to use the previous post in a thread as context.

Key points

  • Features: there are four groups, fed to LibSVM with an RBF kernel and evaluated with 5-fold cross-validation:
    • lexical unigrams, chosen with χ² and Categorical Proportional Difference;
    • pragmatic cues: emoticons, laughter, punctuation and capitals;
    • implicit phrases: 79 verb phrases and 202 noun phrases mined from about 4,000 tweets;
    • five explicit features: polarity flips, longest run of same-polarity words, counts of positive and negative words, and Lingpipe lexical polarity.
  • Data:
    • Tweet-A: 5,208 tweets, 4,170 sarcastic, labelled by hashtags.
    • Tweet-B: 2,278 tweets, 506 sarcastic, the manually labelled set from Riloff et al.
    • Discussion-A: 1,502 posts from the Internet Argument Corpus, 752 sarcastic.
  • Results on Tweet-A and Discussion-A:
    • On Tweet-A, all features together reach F = 0.8876. The unigram baseline scores 0.842, and the authors' reimplementation of the rule-based Riloff method scores at most 0.474.
    • On Discussion-A, F rises from 0.568 to 0.640.
  • Comparison on Tweet-B: the system reaches F = 0.61. Riloff et al. reported 0.51, and the authors' reimplementation of Maynard and Greenwood (2014) reaches 0.41.
  • Prevalence and thread context:
    • Only about 11% of 18,141 sarcastic tweets contain explicit incongruity, so most sarcasm is implicit.
    • Adding the preceding "elicitor" post to forum posts raises precision to 0.705 but lowers recall to 0.274, because only 15% of posts have an elicitor.

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

Cited by

93 papers of our library cite it (of the 140673 whose bibliographies we read); 20 of them have a page here, 73 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.3115/v1/p15-2124

@inproceedings{joshi2015harnessing,
  title = {{Harnessing Context Incongruity for Sarcasm Detection}},
  author = {Joshi, Aditya and Sharma, Vinita and Bhattacharyya, Pushpak},
  booktitle = {Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing (Volume 2: Short Papers)},
  year = {2015},
  doi = {10.3115/v1/p15-2124},
  url = {https://doi.org/10.3115/v1/p15-2124}
}

References

18 references, 6 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Alias-i. 2008. Lingpipe natural language toolkit. find
  2. Francesco Barbieri, Horacio Saggion, and Francesco Ronzano. 2014. Modelling sarcasm in twitter, a novel approach. ACL 2014, page 50. find our card
  3. John D Campbell and Albert N Katz. 2012. Are there necessary conditions for inducing a sense of sarcastic irony? Discourse Processes, 49(6):459–480. find
  4. Paula Carvalho, Luı́s Sarmento, Mário J Silva, and Eugénio de Oliveira. 2009. Clues for detecting irony in user-generated contents: oh...!! it’s so easy;-). In Proceedings of the 1st international CIKM workshop on Topic-sentiment analysis for mass opinion, pages 53–56. ACM. find
  5. Chih-Chung Chang and Chih-Jen Lin. 2011. Libsvm: a library for support vector machines. ACM Transactions on Intelligent Systems and Technology (TIST), 2(3):27. find
  6. Dmitry Davidov, Oren Tsur, and Ari Rappoport. 2010. Semi-supervised recognition of sarcastic sentences in twitter and amazon. In Proceedings of the Fourteenth Conference on Computational Natural Language Learning, pages 107–116. Association for Computational Linguistics. find our card
  7. Jodi Eisterhold, Salvatore Attardo, and Diana Boxer. 2006. Reactions to irony in discourse: Evidence for the least disruption principle. Journal of Pragmatics, 38(8):1239–1256. find
  8. Roberto Gonzalez-Ibanez, Smaranda Muresan, and Nina Wacholder. 2011. Identifying sarcasm in twitter: a closer look. In Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies: short papers-Volume 2, pages 581–586. Association for Computational Linguistics. find our card
  9. Stacey L Ivanko and Penny M Pexman. 2003. Context incongruity and irony processing. Discourse Processes, 35(3):241–279. find
  10. CC Liebrecht, FA Kunneman, and APJ van den Bosch. 2013. The perfect solution for detecting sarcasm in tweets# not. find our card
  11. Diana Maynard and Mark A Greenwood. 2014. Who cares about sarcastic tweets? investigating the impact of sarcasm on sentiment analysis. In Proceedings of LREC. find our card
  12. Ankit Ramteke, Pushpak Bhattacharyya, Akshat Malu, and J Saketha Nath. 2013. Detecting turnarounds in sentiment analysis: Thwarting. In Proceedings of ACL. find
  13. Ellen Riloff, Ashequl Qadir, Prafulla Surve, Lalindra De Silva, Nathan Gilbert, and Ruihong Huang. 2013. Sarcasm as contrast between a positive sentiment and negative situation. In Proceedings of the 2013 Conference on Empirical Methods in Natural Language Processing, pages 704–714. Association for Computational Linguistics. find our card
  14. Joseph Tepperman, David R Traum, and Shrikanth Narayanan. 2006. yeah right : sarcasm recognition for spoken dialogue systems. In INTERSPEECH. find
  15. Oren Tsur, Dmitry Davidov, and Ari Rappoport. 2010a. Icwsm-a great catchy name: Semi-supervised recognition of sarcastic sentences in online product reviews. In ICWSM. find
  16. Oren Tsur, Dmitry Davidov, and Ari Rappoport. 2010b. Icwsm-a great catchy name: Semi-supervised recognition of sarcastic sentences in online product reviews. In ICWSM. find
  17. Marilyn A Walker, Jean E Fox Tree, Pranav Anand, Rob Abbott, and Joseph King. 2012. A corpus for research on deliberation and debate. In LREC, pages 812–817. find
  18. Deirdre Wilson. 2006. The pragmatics of verbal irony: Echo or pretence? Lingua, 116(10):1722–1743. find