Elamur Research⌕
☰
home / papers / 2019 / GECOR: An End-to-End Generative Ellipsis and Co-reference Resolution Model for Task-Oriented Dialogue

GECOR: An End-to-End Generative Ellipsis and Co-reference Resolution Model for Task-Oriented Dialogue

Summary

Task-oriented dialogue systems often misread user turns that leave information out or refer back to earlier content with pronouns, which can derail task completion. GECOR handles ellipsis and co-reference in one generative model. Instead of detecting mentions and ranking candidate antecedents, it rewrites an incomplete user turn into a self-contained version, pulling missing material from the prior dialogue. It is a seq2seq network with two bidirectional GRU encoders, one for the current utterance and one for the conversation history, plus a GRU decoder that mixes vocabulary generation with copying (plain or gated) from the context. The authors also add GECOR as an auxiliary decoder in a multi-task extension of the TSCP end-to-end dialogue system, next to state tracking and response generation. For training, they hand-annotated CamRest676 with recovered ellipsis and resolved anaphora. GECOR beats a seq2seq baseline on EM, BLEU and F1, and the joint model raises task success F1 over TSCP, most clearly on co-reference data. The released dataset and code make it a reusable reference for context-aware utterance rewriting.

Key points

  • Dataset: a new resource built on CamRest676, where omitted expressions are filled in and anaphors are replaced by their antecedents. It is split into ellipsis, co-reference and mixed variants and published with code at multinlp.github.io/GECOR.
  • Model details: 50-dimensional GloVe embeddings, single-layer BiGRU encoders, Bahdanau-style attention, and a copy mechanism following Gu et al. (2016) or a gated variant adapted from See et al. (2017).
  • Resolution quality: GECOR leaves already complete inputs exactly intact in 92.03% of cases. For incomplete inputs, 42.04% of outputs exactly match the reference. Most mismatches are still complete utterances, often paraphrases, with only 6.3% left unresolved.
  • Downstream gains: compared with TSCP, the multi-task model improves success F1 by about 4 points on co-reference data, 2.7 on mixed data and 0.8 on ellipsis data. The co-reference result nearly matches TSCP trained on gold complete utterances.
  • Remaining errors: partial resolution of pronouns that refer to several items, word repetition, and minor substitutions. The authors suggest strengthening the attention and copy components to address them.

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

Topics: Coreference resolution

Cited by

39 papers of our library cite it (of the 140673 whose bibliographies we read); 1 of them have a page here, 38 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/d19-1462 · arXiv 1909.12086

@inproceedings{quan2019gecor,
  title = {{GECOR: An End-to-End Generative Ellipsis and Co-reference Resolution Model for Task-Oriented Dialogue}},
  author = {Quan, Jun and Xiong, Deyi and Webber, Bonnie and Hu, Changjian},
  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/d19-1462},
  eprint = {1909.12086},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.18653/v1/d19-1462}
}

References

28 references, 10 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Dzmitry Bahdanau, Kyunghyun Cho, and Yoshua Bengio. 2015. Neural machine translation by jointly learning to align and translate. Proceedings of the 3rd International Conference on Learning Representations (ICLR 2015). find our card
  2. Kevin Clark and Christopher D Manning. 2016. Deep reinforcement learning for mention-ranking coreference models. Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, pages 2256–2262. find our card
  3. Mary Dalrymple, Stuart M Shieber, and Fernando CN Pereira. 1991. Ellipsis and higher-order unification. Linguistics and philosophy, 14(4):399–452. find
  4. Mihail Eric and Christopher D Manning. 2017b. Key-value retrieval networks for task-oriented dialogue. Proceedings of the SIGDIAL 2017 Conference, pages 37–49. find
  5. Jiatao Gu, Zhengdong Lu, Hang Li, and Victor OK Li. 2016. Incorporating copying mechanism in sequence-to-sequence learning. Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics, pages 1631–1640. find our card
  6. Xisen Jin, Wenqiang Lei, Zhaochun Ren, Hongshen Chen, Shangsong Liang, Yihong Zhao, and Dawei Yin. 2018. Explicit state tracking with semi-supervisionfor neural dialogue generation. In Proceedings of the 27th ACM International Conference on Information and Knowledge Management, pages 1403–1412. ACM. find
  7. Haoruo Peng, Kai-Wei Chang, and Dan Roth. 2015. A joint framework for coreference resolution and mention head detection. In Proceedings of the Nineteenth Conference on Computational Natural Language Learning, pages 12–21. find our card
  8. Diederik P Kingma and Jimmy Ba. 2015. Adam: A method for stochastic optimization. Proceedings of the 3rd International Conference on Learning Representations (ICLR 2015). find
  9. Jeffrey Pennington, Richard Socher, and Christopher Manning. 2014. Glove: Global vectors for word representation. In Proceedings of the 2014 conference on empirical methods in natural language processing (EMNLP), pages 1532–1543. find
  10. Vineet Kumar and Sachindra Joshi. 2016. Nonsentential question resolution using sequence to sequence learning. In Proceedings of COLING 2016, the 26th International Conference on Computational Linguistics: Technical Papers, pages 2022–2031. find
  11. Kenton Lee, Luheng He, Mike Lewis, and Luke Zettlemoyer. 2017. End-to-end neural coreference resolution. Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pages 188–197. find our card
  12. Kenton Lee, Luheng He, and Luke Zettlemoyer. 2018. Higher-order coreference resolution with coarse-to-fine inference. Proceedings of NAACL-HLT 2018, pages 687–692. find our card
  13. Wenqiang Lei, Xisen Jin, Min-Yen Kan, Zhaochun Ren, Xiangnan He, and Dawei Yin. 2018. Sequic-ity: Simplifying task-oriented dialogue systems with single sequence-to-sequence architectures. In Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Pape… find
  14. Jiwei Li, Will Monroe, Alan Ritter, Michel Galley, Jianfeng Gao, and Dan Jurafsky. 2016. Deep reinforcement learning for dialogue generation. Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, pages 1192– 1202. find our card
  15. Zhengzhong Liu, Edgar Gonzàlez Pellicer, and Daniel Gillick. 2016. Exploring the steps of verb phrase ellipsis. In Proceedings of the Workshop on Coreference Resolution Beyond OntoNotes (CORBON 2016), pages 32–40. find our card
  16. Leif Arda Nielsen. 2003. A corpus-based study of verb phrase ellipsis. In Proceedings of the 6th Annual cluk Research Colloquium, pages 109–115. find
  17. Martha S. Palmer, Deborah A. Dahl, Rebecca J. Schiffman, Lynette Hirschman, Marcia Linebarger, and John Dowding. 1986. Recovering implicit information. In Proceedings of the Workshop on Strategic Computing Natural Language, HLT ’86, pages 96– 113, Stroudsburg, PA, USA. Association for Computational… find
  18. Kishore Papineni, Salim Roukos, Todd Ward, and Wei-Jing Zhu. 2002. Bleu: a method for automatic evaluation of machine translation. In Proceedings of the 40th annual meeting on association for computational linguistics, pages 311–318. Association for Computational Linguistics. find our card
  19. Alexander I Rudnicky, Eric Thayer, Paul Constantinides, Chris Tchou, R Shern, Kevin Lenzo, Wei Xu, and Alice Oh. 1999. Creating natural dialogs in the carnegie mellon communicator system. In Sixth European Conference on Speech Communication and Technology. find
  20. Abigail See, Peter J. Liu, and Christopher D. Manning. 2017. Get to the point: Summarization with pointer-generator networks. In Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1073– 1083, Vancouver, Canada. Association for Compu… find
  21. Stuart M Shieber, Fernando CN Pereira, and Mary Dalrymple. 1996. Interactions of scope and ellipsis. Linguistics and philosophy, 19(5):527–552. find
  22. Olga Uryupina and Alessandro Moschitti. 2013. Multilingual mention detection for coreference resolution. In Proceedings of the Sixth International Joint Conference on Natural Language Processing, pages 100–108. find
  23. Oriol Vinyals and Quoc Le. 2015. A neural conversational model. Proceedings of the International Conference on Machine Learning, Deep Learning Workshop (2015). find our card
  24. Tsung-Hsien Wen, Milica Gasic, Nikola Mrkšić, Lina M Rojas Barahona, Pei-Hao Su, Stefan Ultes, David Vandyke, and Steve Young. 2016a. Conditional generation and snapshot learning in neural dialogue systems. In Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing,… find
  25. Tsung-Hsien Wen, David Vandyke, Nikola Mrksic, Milica Gasic, Lina M Rojas-Barahona, Pei-Hao Su, Stefan Ultes, and Steve Young. 2016b. A network-based end-to-end trainable task-oriented dialogue system. In Proceedings of the 15th Conference of the European Chapter of the Association for Computational… find
  26. Jie Zheng, Fang Kong, and Guodong Zhou. 2018. Sequence to sequence model to ellipsis recovery for chinese short text. Journal of Chinese Information Processing, 32(12):92–99. find
  27. Victor Zue, Stephanie Seneff, James R Glass, Joseph Polifroni, Christine Pao, Timothy J Hazen, and Lee Hetherington. 2000. Juplter: a telephone-based conversational interface for weather information. IEEE Transactions on speech and audio processing, 8(1):85–96. find
  28. Victor W Zue and James R Glass. 2000. Conversational interfaces: Advances and challenges. Proceedings of the IEEE, 88(8):1166–1180. find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/papers/quan-2019-gecor-end-end-generative-ellipsis/