Elamur Research⌕
☰
home / papers / 2019 / A Modular Architecture for Unsupervised Sarcasm Generation

A Modular Architecture for Unsupervised Sarcasm Generation

Summary

The paper addresses automatic sarcasm generation: rewriting a literal negative opinion (e.g., a complaint about a late bus) as a sarcastic version, without any parallel training pairs. The approach builds on the context-incongruity theory, which treats sarcasm as a clash between positive sentiment and a negative situation. The authors design a four-module pipeline. First, an LSTM classifier with self-attention strips sentiment-heavy words. Next, an attention-and-copy seq2seq model rewrites the neutral remainder as a strongly positive phrase. A retrieval component then pulls a related negative situation from a gazetteer bootstrapped from sarcastic texts. Finally, a synthesizer tuned with reinforcement learning merges both into a sarcastic sentence. Training needs only separate, non-aligned collections of positive, negative, and sarcastic opinion texts. Since no evaluation data existed, the authors manually built a small benchmark of literal opinions with sarcastic paraphrases. On automatic metrics and human ratings, the system generally produces better sarcasm than unsupervised MT and style-transfer baselines. The work is an early step toward pragmatics-aware, unsupervised text generation, and its code and data are publicly released.

Key points

  • The pipeline has four stages: Sentiment Neutralization, Positive Sentiment Induction, Negative Situation Retrieval, and Sarcasm Synthesis. Neutralization drops words whose attention weight exceeds µ + 2σ.
  • Negative situations are word n-grams (n ≤ 5) that follow positive phrases in sarcastic sentences. They are mined with bootstrapping in the style of Riloff et al. (2013).
  • Baselines include SarcasmBot, Monoses, UNMT, ST, FLIP, and the authors' own simpler variants. SarcasmBot produced only 20 distinct outputs across the whole test set. UNMT and Monoses tended to simply copy the input.
  • Evaluation combined translation and document-similarity metrics with human judgments of sarcasm intensity, fluency, and adequacy. The RL variants produced longer, more context-rich outputs but were less fluent.
  • Most errors trace back to faulty neutralization weights, dropped or spurious words during positive induction, and imprecise sarcasm scoring in RL. Code and data are available at github.com/TarunTater/sarcasm_generation.

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

20 papers of our library cite it (of the 140673 whose bibliographies we read); 7 of them have a page here, 13 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/d19-1636

@inproceedings{mishra2019modular,
  title = {{A Modular Architecture for Unsupervised Sarcasm Generation}},
  author = {Mishra, Abhijit and Tater, Tarun and Sankaranarayanan, Karthik},
  booktitle = {Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP)},
  year = {2019},
  doi = {10.18653/v1/d19-1636},
  url = {https://doi.org/10.18653/v1/d19-1636}
}

References

39 references, 20 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Mikel Artetxe, Gorka Labaka, and Eneko Agirre. 2018. Unsupervised statistical machine translation. arXiv preprint arXiv:1809.01272. ↗ our card
  2. Mikel Artetxe, Gorka Labaka, Eneko Agirre, and Kyunghyun Cho. 2017. Unsupervised neural machine translation. arXiv preprint arXiv:1710.11041. ↗ our card
  3. Dzmitry Bahdanau, Kyunghyun Cho, and Yoshua Bengio. 2014. Neural machine translation by jointly learning to align and translate. arXiv:1409.0473. ↗ our card
  4. Satanjeev Banerjee and Alon Lavie. 2005. Meteor: An automatic metric for mt evaluation with improved correlation with human judgments. In Proceedings of the acl workshop on intrinsic and extrinsic evaluation measures for machine translation and/or summarization, pages 65–72. find our card
  5. Francesco Barbieri, Horacio Saggion, and Francesco Ronzano. 2014. Modelling sarcasm in twitter, a novel approach. ACL 2014, page 50. find our card
  6. Elisabeth Camp. 2012. Sarcasm, pretense, and the semantics/pragmatics distinction. Noûs, 46(4):587– 634. ↗
  7. 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. ↗
  8. 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
  9. Aniruddha Ghosh and Tony Veale. 2016. Fracking sarcasm using neural network. In Proceedings of the 7th workshop on computational approaches to subjectivity, sentiment and social media analysis, pages 161–169. ↗ our card
  10. Roberto González-Ibánez, 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
  11. Caglar Gulcehre, Sungjin Ahn, Ramesh Nallap-ati, Bowen Zhou, and Yoshua Bengio. 2016. Pointing the unknown words. arXiv preprint arXiv:1603.08148. ↗
  12. Sepp Hochreiter and Jürgen Schmidhuber. 1997. Long short-term memory. Neural computation, 9(8):1735–1780. ↗
  13. Abhijit Mishra, Kuntal Dey, and Pushpak Bhattacharyya. 2017. Learning cognitive features from gaze data for sentiment and sarcasm classification using convolutional neural network. In Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), volume 1, pages 377–387. find our card
  14. Eduard Hovy. 1987. Generating natural language under pragmatic constraints. Journal of Pragmatics, 11(6):689–719. ↗
  15. Abhijit Mishra, Diptesh Kanojia, Seema Nagar, Kuntal Dey, and Pushpak Bhattacharyya. 2016. Harnessing cognitive features for sarcasm detection. ACL 2016, page 156. ↗ our card
  16. Parag Jain, Abhijit Mishra, Amar Prakash Azad, and Karthik Sankaranarayanan. 2018. Unsupervised controllable text formalization. arXiv preprint arXiv:1809.04556. ↗
  17. Kishore Papineni, Salim Roukos, Todd Ward, and Wei-Jing Zhu. 2002. Bleu: a method for automatic evaluation of machine translation. In ACL, 2002. find our card
  18. Aditya Joshi, Samarth Agrawal, Pushpak Bhattacharyya, and Mark J Carman. 2017a. Expect the unexpected: Harnessing sentence completion for sarcasm detection. In International Conference of the Pacific Association for Computational Linguistics, pages 275–287. Springer. find
  19. Aditya Joshi, Pushpak Bhattacharyya, and Mark J Carman. 2017b. Automatic sarcasm detection: A survey. ACM Computing Surveys (CSUR), 50(5):73. ↗ our card
  20. Aditya Joshi, Anoop Kunchukuttan, Pushpak Bhattacharyya, and Mark James Carman. 2015a. Sarcasmbot: An open-source sarcasm-generation module for chatbots. In WISDOM Workshop at KDD. find
  21. Aditya Joshi, Vinita Sharma, and Pushpak Bhattacharyya. 2015b. Harnessing context incongruity for sarcasm detection. In 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), volume 2, pages 757–762. find our card
  22. Guillaume Klein, Yoon Kim, Yuntian Deng, Jean Senellart, and Alexander M Rush. 2017. Opennmt: Open-source toolkit for neural machine translation. arXiv preprint arXiv:1701.02810. ↗ our card
  23. Dimitrios Kotzias, Misha Denil, Nando De Freitas, and Padhraic Smyth. 2015. From group to individual labels using deep features. In Proceedings of the 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 597–606. ACM. find
  24. Juncen Li, Robin Jia, He He, and Percy Liang. 2018. Delete, retrieve, generate: A simple approach to sentiment and style transfer. arXiv preprint arXiv:1804.06437. ↗ our card
  25. Chin-Yew Lin. 2004. Rouge: A package for automatic evaluation of summaries. Text Summarization Branches Out. find
  26. Edward Loper and Steven Bird. 2002. Nltk: the natural language toolkit. arXiv preprint cs/0205028. find
  27. Paul Piwek. 2003. A flexible pragmatics-driven language generator for animated agents. arXiv preprint cs/0312050. find
  28. Ellen Riloff, Ashequl Qadir, Prafulla Surve, Lalin-dra 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. find our card
  29. Graeme Ritchie. 2005. Computational mechanisms for pun generation. In Proceedings of the Tenth European Workshop on Natural Language Generation (ENLG-05). find
  30. Tianxiao Shen, Tao Lei, Regina Barzilay, and Tommi Jaakkola. 2017. Style transfer from non-parallel text by cross-alignment. In Advances in neural information processing systems, pages 6830–6841. ↗ our card
  31. Advaith Siddharthan. 2014. A survey of research on text simplification. ITL-International Journal of Applied Linguistics, 165(2):259–298. ↗
  32. Carlo Strapparava, Oliviero Stock, and Rada Mihalcea. 2011. Computational humour. In Emotion-oriented systems, pages 609–634. Springer. find
  33. Mukund Sundararajan, Ankur Taly, and Qiqi Yan. 2017. Axiomatic attribution for deep networks. In Proceedings of the 34th International Conference on Machine Learning-Volume 70, pages 3319–3328. JMLR. org. ↗
  34. Yi Tay, Luu Anh Tuan, Siu Cheung Hui, and Jian Su. 2018. Reasoning with sarcasm by reading in-between. arXiv preprint arXiv:1805.02856. ↗ our card
  35. Byron C Wallace. 2015. Computational irony: A survey and new perspectives. Artificial Intelligence Review, 43(4):467–483. find
  36. Robert West and Eric Horvitz. 2019. Reverse-engineering satire, or” paper on computational humor accepted despite making serious advances”. arXiv preprint arXiv:1901.03253. ↗
  37. Ronald J Williams. 1992. Simple statistical gradient-following algorithms for connectionist reinforcement learning. Machine learning, 8(3-4):229–256. ↗
  38. Jingjing Xu, Xu Sun, Qi Zeng, Xuancheng Ren, Xiaodong Zhang, Houfeng Wang, and Wenjie Li. 2018. Unpaired sentiment-to-sentiment translation: A cycled reinforcement learning approach. arXiv preprint arXiv:1805.05181. ↗ our card
  39. Zhirui Zhang, Shuo Ren, Shujie Liu, Jianyong Wang, Peng Chen, Mu Li, Ming Zhou, and Enhong Chen. 2018. Style transfer as unsupervised machine translation. arXiv preprint arXiv:1808.07894. ↗ our card

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/papers/mishra-2019-modular-architecture-unsupervised-sarcasm-generation/