Elamur Research⌕
☰
home / papers / 2022 / Reflect, Not Reflex: Inference-Based Common Ground Improves Dialogue Response Quality

Reflect, Not Reflex: Inference-Based Common Ground Improves Dialogue Response Quality

Summary

Dialogue response generation (RG) models tend to produce bland, generic replies. The authors argue this happens because the models answer "reflexively": they map dialogue history straight to a response and skip the step people take of building common ground (CG), meaning the shared knowledge and beliefs between speakers. They treat CG as commonsense inferences that one participant makes about the other, written as natural-language question–answer pairs.

They built Reflect through two crowdsourcing stages on Amazon Mechanical Turk. First, workers answered inference questions about a dialogue context. Then other workers wrote a reply grounded in each inference. Human evaluation shows that Reflect responses are more often sensible, specific and interesting than both original dataset responses and the output of models trained on such data.

Using CG, even in simple ways, also helps models. Few-shot prompts to GPT3 that add an inference question raise the share of high-quality responses by about 30%. The work offers a resource and evidence for building CG-aware dialogue systems.

Key points

  • Reflect contains about 9k human-written responses for 600 dialogue contexts, grounded in 5 inference dimensions. Examples of dimensions are events before or after, speaker emotions, and descriptions of a speaker.
  • Under human judgment, about 58% of Reflect responses are high quality. Original human-written dialogue responses reach 49%, and models trained on reflex-style data score lower still.
  • GPT3-175B few-shot prompting with inference questions (GPT3-FS-InfQ) reaches 54% quality responses, roughly 30% more than vanilla few-shot prompting. Adding the human inference answers as well brings further gains: +16.7% quality and +15% specificity.
  • Fine-tuning BlenderBot (~400M parameters) on Reflect yields about 50% more specific and 26% more high-quality responses. Adding inference questions brings no further gain, which suggests model size matters.
  • Results vary by dimension. GPT3 outperforms Reflect's human responses on event-based dimensions and on the other speaker's emotions, but does worse on "How would you describe A".

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

23 papers of our library cite it (of the 140673 whose bibliographies we read); 2 of them have a page here, 21 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2022.emnlp-main.714 · arXiv 2211.09267

@inproceedings{zhou2022reflect,
  title = {{Reflect, Not Reflex: Inference-Based Common Ground Improves Dialogue Response Quality}},
  author = {Zhou, Pei and Cho, Hyundong and Jandaghi, Pegah and Lee, Dong-Ho and Lin, Bill Yuchen and Pujara, Jay and Ren, Xiang},
  booktitle = {Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing},
  year = {2022},
  doi = {10.18653/v1/2022.emnlp-main.714},
  eprint = {2211.09267},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.18653/v1/2022.emnlp-main.714}
}

References

36 references, 7 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Leonard Adolphs, Kurt Shuster, Jack Urbanek, Arthur Szlam, and Jason Weston. 2021. Reason first, then respond: Modular generation for knowledge-infused dialogue. arXiv preprint arXiv:2111.05204. ↗
  2. Jens Allwood. 1976. Linguistic communication as action and cooperation. University of Göteborg. Department of Linguistics. find
  3. David Bohm, Peter M Senge, and Lee Nichol. 2004. On dialogue. Routledge. find
  4. Tom B Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al. 2020. Language models are few-shot learners. arXiv preprint arXiv:2005.14165. ↗
  5. Jerome S Bruner. 1961. The act of discovery. Harvard educational review. find
  6. Hyundong Cho and Jonathan May. 2020. Grounding conversations with improvised dialogues. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 2398–2413, Online. Association for Computational Linguistics. find
  7. Herbert H Clark and Susan E Brennan. 1991. Grounding in communication. ↗
  8. Herbert H Clark and Edward F Schaefer. 1989. Contributing to discourse. Cognitive science, 13(2):259– 294. ↗
  9. Leyang Cui, Yu Wu, Shujie Liu, Yue Zhang, and Ming Zhou. 2020. MuTual: A dataset for multi-turn dialogue reasoning. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 1406–1416, Online. Association for Computational Linguistics. find
  10. Jonathan St BT Evans. 2003. In two minds: dual-process accounts of reasoning. Trends in cognitive sciences, 7(10):454–459. find
  11. Joseph L Fleiss. 1971. Measuring nominal scale agreement among many raters. Psychological bulletin, 76(5):378. ↗
  12. Deepanway Ghosal, Siqi Shen, Navonil Majumder, Rada Mihalcea, and Soujanya Poria. 2022. Cicero: A dataset for contextualized commonsense inference in dialogues. arXiv preprint arXiv:2203.13926. ↗
  13. Herbert P Grice. 1975. Logic and conversation. In Speech acts, pages 41–58. Brill. ↗
  14. Jürgen Habermas. 1985. The theory of communicative action: Volume 1: Reason and the rationalization of society, volume 1. Beacon press. find
  15. Ari Holtzman, Jan Buys, Li Du, Maxwell Forbes, and Yejin Choi. 2019. The curious case of neural text degeneration. In International Conference on Learning Representations. ↗
  16. Jena D. Hwang, Chandra Bhagavatula, Ronan Le Bras, Jeff Da, Keisuke Sakaguchi, Antoine Bosselut, and Yejin Choi. 2021. Comet-atomic 2020: On symbolic and neural commonsense knowledge graphs. In AAAI. ↗ our card
  17. Daniel Kahneman. 2011. Macmillan. Thinking, fast and slow. find
  18. Yanran Li, Hui Su, Xiaoyu Shen, Wenjie Li, Ziqiang Cao, and Shuzi Niu. 2017. DailyDialog: A manually labelled multi-turn dialogue dataset. In Proceedings of the Eighth International Joint Conference on Natural Language Processing (Volume 1: Long Papers), pages 986–995, Taipei, Taiwan. Asian Federation of Natural Language Processing. ↗ our card
  19. A. H. Miller, W. Feng, A. Fisch, J. Lu, D. Batra, A. Bordes, D. Parikh, and J. Weston. 2017. Parlai: A dialog research software platform. arXiv preprint arXiv:1705.06476. ↗ our card
  20. Hannah Rashkin, Eric Michael Smith, Margaret Li, and Y-Lan Boureau. 2019. Towards empathetic open-domain conversation models: A new benchmark and dataset. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 5370–5381, Florence, Italy. Association for Computational Linguistics. find
  21. Nils Reimers and Iryna Gurevych. 2019. Sentence-bert: Sentence embeddings using siamese bert-networks. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing. Association for Computational Linguistics. find
  22. Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Eric Michael Smith, Y-Lan Boureau, and Jason Weston. 2021. Recipes for building an open-domain chatbot. In Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume, pages 300–325, Online. Association for Computational Linguistics. find our card
  23. Maarten Sap, Ronan Le Bras, Emily Allaway, Chandra Bhagavatula, Nicholas Lourie, Hannah Rashkin, Brendan Roof, Noah A. Smith, and Yejin Choi. 2019a. ATOMIC: an atlas of machine commonsense for if-then reasoning. In The Thirty-Third AAAI Conference on Artificial Intelligence, AAAI 2019, The Thirty-First Innovative Applications of Artificial Intelligence Conference, IAAI 2019, The Ninth AAAI Symposium on Educational Advances in Artificial Intelligence, EAAI 2019, Honolulu, Hawaii, USA, January 27 - February 1, 2019, pages 3027–3035. AAAI Press. ↗
  24. Maarten Sap, Hannah Rashkin, Derek Chen, Ronan Le Bras, and Yejin Choi. 2019b. Social IQa: Commonsense reasoning about social interactions. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pages 4463– 4473, Hong Kong, China. Association for Computational Linguistics. find
  25. Iulian Serban, Alessandro Sordoni, Ryan Lowe, Laurent Charlin, Joelle Pineau, Aaron Courville, and Yoshua Bengio. 2017. A hierarchical latent variable encoder-decoder model for generating dialogues. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 31. ↗ our card
  26. Kurt Shuster, Mojtaba Komeili, Leonard Adolphs, Stephen Roller, Arthur Szlam, and Jason Weston. 2022. Language models that seek for knowledge: Modular search & generation for dialogue and prompt completion. arXiv preprint arXiv:2203.13224. ↗
  27. Vered Shwartz, Peter West, Ronan Le Bras, Chandra Bhagavatula, and Yejin Choi. 2020. Unsupervised commonsense question answering with self-talk. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 4615–4629. find
  28. Robyn Speer, Joshua Chin, and Catherine Havasi. 2017. Conceptnet 5.5: An open multilingual graph of general knowledge. In Thirty-first AAAI conference on artificial intelligence. ↗
  29. Robert C Stalnaker. 1978. Assertion. In Pragmatics, pages 315–332. Brill. find
  30. Keith E Stanovich and Richard F West. 2000. Individual differences in reasoning: Implications for the rationality debate? Behavioral and brain sciences, 23(5):645–665. ↗
  31. Romal Thoppilan, Daniel De Freitas, Jamie Hall, Noam Shazeer, Apoorv Kulshreshtha, Heng-Tze Cheng, Alicia Jin, Taylor Bos, Leslie Baker, Yu Du, et al. 2022. Lamda: Language models for dialog applications. arXiv preprint arXiv:2201.08239. ↗ our card
  32. Gordon Wells. 2000. Dialogic inquiry in education. Vygotskian perspectives on literacy research, pages 51–85. find
  33. Yizhe Zhang, Siqi Sun, Michel Galley, Yen-Chun Chen, Chris Brockett, Xiang Gao, Jianfeng Gao, Jingjing Liu, and Bill Dolan. 2020. DIALOGPT : Large-scale generative pre-training for conversational response generation. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics: System Demonstrations, pages 270–278, Online. Association for Computational Linguistics. ↗ our card
  34. Tiancheng Zhao, Ran Zhao, and Maxine Eskenazi. 2017. Learning discourse-level diversity for neural dialog models using conditional variational autoencoders. In Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 654–664. find
  35. Pei Zhou, Karthik Gopalakrishnan, Behnam Hedayatnia, Seokhwan Kim, Jay Pujara, Xiang Ren, Yang Liu, and Dilek Hakkani-Tur. 2021. Commonsense-focused dialogues for response generation: An empirical study. In Proceedings of the 22nd Annual Meeting of the Special Interest Group on Discourse and Dialogue, pages 121–132, Singapore and Online. Association for Computational Linguistics. ↗
  36. Pei Zhou, Karthik Gopalakrishnan, Behnam Hedayatnia, Seokhwan Kim, Jay Pujara, Xiang Ren, Yang Liu, and Dilek Hakkani-Tur. 2022. Think before you speak: Using self-talk to generate implicit commonsense knowledge for response generation. find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/papers/zhou-2022-reflect-not-reflex-inference-based/