Elamur Research⌕
☰
home / papers / 2019 / Exploring Author Context for Detecting Intended vs Perceived Sarcasm

Exploring Author Context for Detecting Intended vs Perceived Sarcasm

Summary

The paper asks whether a tweet author's posting history helps decide if that tweet is sarcastic, and whether the answer depends on how the sarcasm labels were produced. The authors represent each user as an embedding built from tweets posted before the target tweet. They use four embedding models: CASCADE, a recency-weighted variant they propose (W-CASCADE), an encoder-decoder model (ED) and a SUMMARY model. These embeddings are used either alone ("exclusive" models) or concatenated with text features from the SIARN baseline ("inclusive" models). Experiments cover two Twitter corpora: Riloff, annotated by humans, and Ptacek, labelled through sarcasm hashtags.

On Ptacek, user history alone outperforms the text-only state of the art, and combined models set a new best score. On Riloff, history-only models perform poorly and the gains from adding history are marginal. The authors argue that hashtag labels capture sarcasm the author intends, while manual labels capture sarcasm readers perceive. They conclude that detection research should treat these as distinct phenomena.

Key points

  • Only part of each dataset could be re-collected with user timelines: 701 of 3,200 Riloff tweets and 27,177 of 50,000 Ptacek tweets. Data splits were stratified by user, so no user's embedding appears in more than one split.
  • On Ptacek, EX-W-CASCADE uses no tweet content yet reaches F1 0.922, versus 0.863 for SIARN. IN-W-CASCADE gives the best result at 0.934.
  • On Riloff, the best model is IN-ED with F1 0.739, versus 0.711 for SIARN. Exclusive models stay at or below 0.546.
  • Hashtags and manual labels disagree often in Riloff: 217 of the 509 tweets annotated as non-sarcastic contain a sarcasm tag.
  • The authors re-labelled Riloff by hashtags to create #Riloff. On #Riloff, exclusive models reach up to 0.827 F1, which supports the intended-vs-perceived distinction.

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

30 papers of our library cite it (of the 140673 whose bibliographies we read); 10 of them have a page here, 20 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/p19-1275 · arXiv 1910.11932

@inproceedings{oprea2019exploring,
  title = {{Exploring Author Context for Detecting Intended vs Perceived Sarcasm}},
  author = {Oprea, Silviu and Magdy, Walid},
  booktitle = {Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics},
  year = {2019},
  doi = {10.18653/v1/p19-1275},
  eprint = {1910.11932},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.18653/v1/p19-1275}
}

References

26 references, 9 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Amir, S.; Wallace, B. C.; Lyu, H.; Carvalho, P.; and Silva, M. J. 2016. Modelling context with user embeddings for sarcasm detection in social media. In CoNLL, 167–177. ACL. find our card
  2. Bamman, D., and Smith, N. A. 2015. Contextualized sarcasm detection on twitter. In ICWSM, 574–577. AAAI Press. find
  3. Campbell, J. D., and Katz, A. N. 2012. Are there necessary conditions for inducing a sense of sarcastic irony? Discourse Processes 49(6):459–480. ↗
  4. Dress, M. L.; Kreuz, R. J.; Link, K. E.; and Caucci, G. M. 2008. Regional variation in the use of sarcasm. JLS 27(1):71–85. ↗
  5. Goldberg, L. R. 1993. The structure of phenotypic personality traits. American psychologist 48(1):26. ↗
  6. Hazarika, D.; Poria, S.; Gorantla, S.; Cambria, E.; Zimmermann, R.; and Mihalcea, R. 2018. Cascade: Contextual sarcasm detection in online discussion forums. In COLING, 1837–1848. ACL. find our card
  7. Rajadesingan, A.; Zafarani, R.; and Liu, H. 2015. Sarcasm detection on twitter: A behavioral modeling approach. In WSDM, 97–106. ACM. find
  8. Riloff, E.; Qadir, A.; Surve, P.; De Silva, L.; Gilbert, N.; and Huang, R. 2013. Sarcasm as contrast between a positive sentiment and negative situation. In EMNLP, 704–714. ACL. find our card
  9. Rockwell, P., and Theriot, E. M. 2001a. Culture, gender, and gender mix in encoders of sarcasm: A self-assessment analysis. Communication Research Reports 18(1):44–52. ↗
  10. Rockwell, P., and Theriot, E. M. 2001b. Culture, gender, and gender mix in encoders of sarcasm: A self-assessment analysis. Communication Research Reports 18(1):44–52. ↗
  11. Schuster, M., and Paliwal, K. K. 1997. Bidirectional recurrent neural networks. IEEE Transactions on Signal Processing 45(11):2673–2681. ↗
  12. Sutskever, I.; Vinyals, O.; and Le, Q. V. 2014. Sequence to sequence learning with neural networks. In NIPS, 3104–3112. MIT Press. ↗
  13. Joshi, A.; Tripathi, V.; Patel, K.; Bhattacharyya, P.; and Carman, M. 2016. Are word embedding-based features useful for sarcasm detection? In EMNLP, 1006–1011. ACL. ↗ our card
  14. Tay, Y.; Luu, A. T.; Hui, S. C.; and Su, J. 2018. Reasoning with sarcasm by reading in-between. In ACL, 1010–1020. ACL. find our card
  15. Kellogg, R. T. 2001. Long-term working memory in text production. Memory & Cognition 29(1):43–52. ↗
  16. Tieleman, T., and Hinton, G. 2012. Lecture 6.5— RMSProp: Divide the gradient by a running average of its recent magnitude. Coursera: Neural Networks for Machine Learning. Accessed: 2019-03-01. find
  17. Klein, G.; Kim, Y.; Deng, Y.; Senellart, J.; and Rush, A. 2017. OpenNMT: Open-source toolkit for neural machine translation. In ACL, 67–72. ACL. find our card
  18. Le, Q., and Mikolov, T. 2014. Distributed Representations of Sentences and Documents. In ICML, 1188– 1196. PMLR. ↗ our card
  19. Luong, T.; Pham, H.; and Manning, C. D. 2015. Effective approaches to attention-based neural machine translation. In EMNLP, 1412–1421. ACL. ↗
  20. Wallace, B. C.; Choe, D. K.; and Charniak, E. 2015. Sparse, contextually informed models for irony detection: Exploiting user communities, entities and sentiment. In ACL-IJCNLP, 1035–1044. ACL. find
  21. Wilson, D. 2006. The pragmatics of verbal irony: Echo or pretence? Lingua 116(10):1722–1743. ↗
  22. Matthews, G., and Gilliland, K. 1999. The personality theories of H.J. Eysenck and J.A. Gray: a comparative review. Personality and Individual Differences 26(4):583–626. find
  23. Maynard, D., and Greenwood, M. 2014. Who cares about sarcastic tweets? investigating the impact of sarcasm on sentiment analysis. In LREC. ELRA. find our card
  24. Pennington, J.; Socher, R.; and Manning, C. 2014. Glove: Global vectors for word representation. In EMNLP, 1532–1543. ACL. find
  25. Pexman, P. M. 2005. Social Factors in the Interpretation of Verbal Irony: The Roles of Speaker and Listener Characteristics. find
  26. Ptáček, T.; Habernal, I.; and Hong, J. 2014. Sarcasm detection on czech and english twitter. In COLING, 213–223. ACL. find our card

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/oprea-2019-exploring-author-context-detecting-intended/