Sarcasm as Contrast between a Positive Sentiment and Negative Situation

What it is about

The paper looks at one common form of sarcasm on Twitter: a positive sentiment verb or predicate placed just before a stereotypically negative situation. The authors learn both lists by bootstrapping from a single seed word, "love" (p. 2), using tweets with a #sarcasm hashtag as positives. Candidate phrases are scored by the share of sarcastic tweets among tweets that contain them, with a frequency floor and a threshold of 0.80 for situations and 0.70 for sentiments. The data is about 35,000 hashtagged tweets and 140,000 random tweets. The learned lists hold 26 positive verb phrases, 20 predicative expressions and 239 negative situations. On a 3,000-tweet test set the ordered contrast rule (positive within 5 words before a negative situation) reaches 70% precision at about 9% recall. The hybrid with an SVM raises the F score from 48% to 51%, and the paper reports that the gain is significant at p < .01. Unordered sentiment pairs from standard lexicons do much worse, which shows that the order of the two parts matters.

Ideas

• I1. A contrast rule in prag/src/irony.rs. The module already flags irony by incongruity. Add the ordered pattern "positive sentiment, then a negative situation within 5 tokens" as a second, explained cue, with the matched words in the explanation. Test on Mova's own open text: use the teacher to label sarcastic lines in public-domain dialogue, then check precision and recall by two-fold cv. Do not use the Twitter data for training.
• I2. Bootstrapping with a deterministic loop in Rust. The algorithm is seed, harvest the n-grams on one side of the known term, score by a conditional rate, keep the top few above a threshold, and repeat until nothing new passes. Implement it with BTreeMap ordering and fixed tie-breaks, and run it twice to show the lists are identical. Use it to grow the lists of "pragmatics expressions" in Expressions with their real meanings — the pragmatics lens, with a frequency floor and subsumption removal (a shorter phrase replaces a longer one that it covers).
• I3. Negative situations as world knowledge. The paper's main point is that the negative part is a cultural norm, not a word with negative polarity. Store situations such as queueing for a long time, being left out of a game, or a visit to the dentist as default negative states in onto (concepts with defaults), so that the world reader can say why a happy word sits in a bad event. Measure on the feeling questions of the tale readers in world/src/ftqa.rs: does the default raise the share of correct "how did X feel" answers when the text says the event but not the feeling?
• I4. Ablation for the claim. Run the same lists with the order reversed and without the proximity limit, and report the precision and recall of each variant against the teacher labels. If the ordered rule is not better than the unordered one, the claim does not hold for Mova's data, and the card is closed.

For Mova

Chapters: 34. Irony and figurative language (the irony section) and 33. Implicature and presupposition. Code: prag/src/irony.rs (already cites the line of work), prag/src/lens.rs and prag/src/feats.rs for features, onto for defaults. Doc: Pragmatics: algorithms from the literature, for Mova (2026-10-03). Licence note: the tweets are CC BY-NC-SA data, so they may serve as a test at most, and never as training data or as a list in a release. The paper's learned phrase lists are not to be copied into the repo. Mova's lists come from its own runs on open text.

Cite

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

@inproceedings{riloff2013sarcasm,
  title = {{Sarcasm as Contrast between a Positive Sentiment and Negative Situation}},
  author = {Riloff, Ellen and Qadir, Ashequl and Surve, Prafulla and Silva, Lalindra De and Gilbert, Nathan and Huang, Ruihong},
  booktitle = {Proceedings of the 2013 Conference on Empirical Methods in Natural Language Processing},
  year = {2013},
  url = {https://aclanthology.org/D13-1066/}
}

References

22 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Taylor Berg-Kirkpatrick, David Burkett, and Dan Klein. 2012. An empirical investigation of statistical significance in nlp. In Proceedings of the 2012 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning, EMNLP-CoNLL ’12, pages 995–1005. find
2. 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, TSA 2009. find
3. Chih-Chung Chang and Chih-Jen Lin. 2011. LIBSVM: A library for support vector machines. ACM Transactions on Intelligent Systems and Technology, 2:27:1– 27:27. find
4. Henry S. Cheang and Marc D. Pell. 2008. The sound of sarcasm. Speech Commun., 50(5):366–381, May. find
5. Henry S. Cheang and Marc D. Pell. 2009. Acoustic markers of sarcasm in cantonese and english. The Journal of the Acoustical Society of America, 126(3):1394–1405. 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, CoNLL 2010. find
7. Elena Filatova. 2012. Irony and sarcasm: Corpus generation and analysis using crowdsourcing. In Proceedings of the Eight International Conference on Language Resources and Evaluation (LREC’12). find
8. Roger J. Kreuz Gina M. Caucci. 2012. Social and paralinguistic cues to sarcasm. online 08/02/2012, 25:1– 22, February. find
9. Roberto González-Ibáñez, 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. find
10. Lars Kai Hansen, Adam Arvidsson, Finn Arup Nielsen, Elanor Colleoni, and Michael Etter. 2011. Good friends, bad news - affect and virality in twitter. In The 2011 International Workshop on Social Computing, Network, and Services (SocialComNet 2011). find
11. Roger Kreuz and Gina Caucci. 2007. Lexical influences on the perception of sarcasm. In Proceedings of the Workshop on Computational Approaches to Figurative Language. find
12. Christine Liebrecht, Florian Kunneman, and Antal Van den Bosch. 2013. The perfect solution for detecting sarcasm in tweets #not. In Proceedings of the 4th Workshop on Computational Approaches to Subjectivity, Sentiment and Social Media Analysis, WASSA 2013. find
13. Bing Liu, Minqing Hu, and Junsheng Cheng. 2005. Opinion observer: Analyzing and comparing opinions on the web. In Proceedings of the 14th International World Wide Web conference (WWW-2005). find
14. Stephanie Lukin and Marilyn Walker. 2013. Really? well. apparently bootstrapping improves the performance of sarcasm and nastiness classifiers for online dialogue. In Proceedings of the Workshop on Language Analysis in Social Media. find
15. Finn Arup Nielsen. 2011. A new anew: Evaluation of a word list for sentiment analysis in microblogs. In Proceedings of the ESWC2011 Workshop on ’Making Sense of Microposts’: Big things come in small packages (http://arxiv.org/abs/1103.2903). ↗
16. Olutobi Owoputi, Brendan O’Connor, Chris Dyer, Kevin Gimpel, Nathan Schneider, and Noah A. Smith. 2013. Improved part-of-speech tagging for online conversational text with word clusters. In The 2013 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL 2013). find
17. Katherine P. Rankin, Andrea Salazar, Maria Luisa Gorno-Tempini, Marc Sollberger, Stephen M. Wilson, Danijela Pavlic, Christine M. Stanley, Shenly Glenn, Michael W. Weiner, and Bruce L. Miller. 2009. Detecting sarcasm from paralinguistic cues: Anatomic and cognitive correlates in neurodegenerative disease. Neuroimage, 47:2005–2015. find
18. Joseph Tepperman, David Traum, and Shrikanth Narayanan. 2006. ”Yeah right”: Sarcasm recognition for spoken dialogue systems. In Proceedings of the INTERSPEECH 2006 - ICSLP, Ninth International Conference on Spoken Language Processing. find
19. Oren Tsur, Dmitry Davidov, and Ari Rappoport. 2010. ICWSM - A Great Catchy Name: Semi-Supervised Recognition of Sarcastic Sentences in Online Product Reviews. In Proceedings of the Fourth International Conference on Weblogs and Social Media (ICWSM-2010), ICWSM 2010. find
20. Michael Wiegand, Josef Ruppenhofer, and Dietrich Klakow. 2013. Predicative adjectives: An unsupervised criterion to extract subjective adjectives. In Proceedings of the 2013 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 534– 539, Atlanta, Georgia, June. Association for Computational Linguistics. find
21. T. Wilson, P. Hoffmann, S. Somasundaran, J. Kessler, J. Wiebe, Y. Choi, C. Cardie, E. Riloff, and S. Patwardhan. 2005a. OpinionFinder: A System for Subjectivity Analysis. In Proceedings of HLT/EMNLP 2005 Interactive Demonstrations, pages 34–35, Vancouver, Canada, October. find
22. Theresa Wilson, Janyce Wiebe, and Paul Hoffmann. 2005b. Recognizing contextual polarity in phrase-level sentiment analysis. In Proceedings of the 2005 Human Language Technology Conference / Conference on Empirical Methods in Natural Language Processing. find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/prag-2013-riloff-sarcasm-as-contrast-between-positive/