Elamur Research⌕
☰
home / papers / 2013 / Leveraging Synthetic Discourse Data via Multi-task Learning for Implicit Discourse Relation Recognition

Leveraging Synthetic Discourse Data via Multi-task Learning for Implicit Discourse Relation Recognition

Summary

This ACL 2013 paper addresses implicit discourse relation recognition: labelling the relation between two text spans when no connective such as but signals it. Annotated implicit data is scarce. Earlier work created synthetic examples by deleting explicit connectives, but Sporleder and Lascarides (2008) found such data transfers poorly to genuine implicit relations. The authors argue that the problem lies in how synthetic data is used, not in the data itself. They build a multi-task learning system based on alternating structure optimization (ASO). The main task is classifying genuine implicit relations in Penn Discourse Treebank 2.0 (PDTB). The auxiliary tasks predict explicit connectives from synthetic data, which comes from two sources: the raw BLLIP news corpus and connective-stripped PDTB explicit relations. Sharing structure across tasks lets the model draw on what explicit and implicit examples have in common while limiting noise. On top-level PDTB relations, multi-task models beat single-task baselines by 5.86% averaged F1. This shows that cheap, noisy synthetic data can help when it is used in the right way.

Key points

  • The method uses ASO-based multi-task learning, with genuine PDTB implicit relations as the main task and explicit connective prediction on synthetic data as auxiliary tasks.
  • The best system improves F1 over a state-of-the-art baseline by:
    • 4.26% for Comparison,
    • 7.06% for Contingency,
    • 8.8% for Expansion,
    • 3.34% for Temporal.
  • Contrary to the authors' expectation, auxiliary data from manually annotated PDTB explicit relations did not outperform noisy BLLIP-derived data. This suggests that costly annotation may not be needed for generating auxiliary data.
  • Absolute performance remains modest, at 44.64% F1. The authors attribute this to two kinds of ambiguity:
    • Implicit relations often allow more than one reading.
    • Connectives are ambiguous. For example, since occurs 184 times as an explicit connective in PDTB: 80 times as Temporal, 94 as Contingency, and 10 as both.

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: Discourse analysis

Cited by

22 papers of our library cite it (of the 140673 whose bibliographies we read); 12 of them have a page here, 10 not yet.

Cite

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

@inproceedings{lan2013leveraging,
  title = {{Leveraging Synthetic Discourse Data via Multi-task Learning for Implicit Discourse Relation Recognition}},
  author = {Lan, Man and Xu, Yu and Niu, Zhengyu},
  booktitle = {Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2013},
  url = {https://aclanthology.org/P13-1047/}
}

References

32 references, 13 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. R.K. Ando and T. Zhang. 2005a. A framework for learning predictive structures from multiple tasks and unlabeled data. The Journal of Machine Learning Research, 6:1817–1853. find
  2. R.K. Ando and T. Zhang. 2005b. A high-performance semi-supervised learning method for text chunking. pages 1–9. Association for Computational Linguistics. Proceedings of the 43rd Annual Meeting on Association for Computational Linguistics. find
  3. A. Argyriou, C.A. Micchelli, M. Pontil, and Y. Ying. 2008. A spectral regularization framework for multi-task structure learning. Advances in Neural Information Processing Systems, 20:2532. find
  4. J. Baxter. 2000. A model of inductive bias learning. J. Artif. Intell. Res. (JAIR), 12:149–198. find
  5. S.J. Blair-Goldensohn. 2007. Long-answer question answering and rhetorical-semantic relations. Ph.D. thesis. find
  6. E. Bonilla, K.M. Chai, and C. Williams. 2008. Multi-task gaussian process prediction. Advances in Neural Information Processing Systems, 20(October). find
  7. L. Carlson, D. Marcu, and M.E. Okurowski. 2001. Building a discourse-tagged corpus in the framework of rhetorical structure theory. pages 1–10. Association for Computational Linguistics. Proceedings of the Second SIGdial Workshop on Discourse and Dialogue-Volume 16. find our card
  8. R. Caruana. 1997. Multitask learning. Learning, 28(1):41–75. find
  9. Machine P. Cimiano, U. Reyle, and J. Saric. 2005. Ontology-driven discourse analysis for information extraction. Data and Knowledge Engineering, 55(1):59–83. find
  10. Eugene Charniak David McClosky and Mark Johnson. 2008. Bllip north american news text, complete. find
  11. T. Evgeniou and M. Pontil. 2004. Regularized multi– task learning. pages 109–117. ACM. Proceedings of the tenth ACM SIGKDD international conference on Knowledge discovery and data mining. find
  12. H. Hernault, D. Bollegala, and M. Ishizuka. 2010. A semi-supervised approach to improve classification of infrequent discourse relations using feature vector extension. pages 399–409. Association for Computational Linguistics. Proceedings of the 2010 Conference on Empirical Methods in Natural Langua… find our card
  13. H. Hernault, D. Bollegala, and M. Ishizuka. 2011. Semi-supervised discourse relation classification with structural learning. In Proceedings of the 12th international conference on Computational linguistics and intelligent text processing - Volume Part I, CICLing’11, pages 340–352, Berlin, Heidelber… find
  14. T. Jebara. 2004. Multi-task feature and kernel selection for svms. page 55. ACM. Proceedings of the twenty-first international conference on Machine learning. find
  15. B. Levin. 1993. English verb classes and alternations: A preliminary investigation, volume 348. University of Chicago press Chicago, IL:. find
  16. Z. Lin, M.Y. Kan, and H.T. Ng. 2009. Recognizing implicit discourse relations in the penn discourse treebank. pages 343–351. Association for Computational Linguistics. Proceedings of the 2009 Conference on Empirical Methods in Natural Language Processing: Volume 1-Volume 1. find our card
  17. W.C. Mann and S.A. Thompson. 1988. Rhetorical structure theory: Toward a functional theory of text organization. Text-Interdisciplinary Journal for the Study of Discourse, 8(3):243–281. find our card
  18. D. Marcu and A. Echihabi. 2002. An unsupervised approach to recognizing discourse relations. pages 368–375. Association for Computational Linguistics. Proceedings of the 40th Annual Meeting on Association for Computational Linguistics. find our card
  19. PDTB-Group. 2008. The penn discourse treebank 2.0 annotation manual. Technical report, Institute for Research in Cognitive Science, University of Pennsylvania. find
  20. E. Pitler, M. Raghupathy, H. Mehta, A. Nenkova, A. Lee, and A. Joshi. 2008. Easily identifiable discourse relations. Citeseer. Proceedings of the 22nd International Conference on Computational Linguistics (COLING 2008), Manchester, UK, August. find our card
  21. E. Pitler, A. Louis, and A. Nenkova. 2009. Automatic sense prediction for implicit discourse relations in text. pages 683–691. Association for Computational Linguistics. Proceedings of the Joint Conference of the 47th Annual Meeting of the ACL and the 4th International Joint Conference on Natural La… find our card
  22. Rashmi Prasad, Nikhil Dinesh, Alan Lee, Eleni Miltsakaki, Livio Robaldo, Aravind Joshi, and Bonnie Webber. 2008. The penn discourse treebank 2.0. In In Proceedings of LREC. find our card
  23. M. Saito, K. Yamamoto, and S. Sekine. 2006. Using phrasal patterns to identify discourse relations. pages 133–136. Association for Computational Linguistics. Proceedings of the Human Language Technology Conference of the NAACL, Companion Volume: Short Papers on XX. find our card
  24. R. Soricut and D. Marcu. 2003. Sentence level discourse parsing using syntactic and lexical information. pages 149–156. Association for Computational Linguistics. Proceedings of the 2003 Conference of the North American Chapter of the Association for Computational Linguistics on Human Language Techn… find our card
  25. C. Sporleder and A. Lascarides. 2008. Using automatically labelled examples to classify rhetorical relations: An assessment. Natural Language Engineering, 14(03):369–416. find
  26. S. Thrun. 1996. Is learning the n-th thing any easier than learning the first? Advances in Neural Information Processing Systems, pages 640–646. find
  27. S. Verberne, L. Boves, N. Oostdijk, and P.A. Coppen. 2007. Evaluating discourse-based answer extraction for why-question answering. pages 735–736. ACM. Proceedings of the 30th annual international ACM SIGIR conference on Research and development in information retrieval. find
  28. W.T. Wang, J. Su, and C.L. Tan. 2010. Kernel based discourse relation recognition with temporal ordering information. pages 710–719. Association for Computational Linguistics. Proceedings of the 48th Annual Meeting of the Association for Computational Linguistics. find our card
  29. B. Wellner, J. Pustejovsky, C. Havasi, A. Rumshisky, and R. Sauri. 2006. Classification of discourse coherence relations: An exploratory study using multiple knowledge sources. pages 117–125. Association for Computational Linguistics. Proceedings of the 7th SIGdial Workshop on Discourse and Dialogue… find our card
  30. T. Wilson, J. Wiebe, and P. Hoffmann. 2009. Recognizing contextual polarity: An exploration of features for phrase-level sentiment analysis. Computational Linguistics, 35(3):399–433. find
  31. F. Wolf, E. Gibson, A. Fisher, and M. Knight. 2005. The discourse graphbank: A database of texts annotated with coherence relations. Linguistic Data Consortium. find
  32. Z.M. Zhou, Y. Xu, Z.Y. Niu, M. Lan, J. Su, and C.L. Tan. 2010. Predicting discourse connectives for implicit discourse relation recognition. pages 1507– 1514. Association for Computational Linguistics. Proceedings of the 23rd International Conference on Computational Linguistics: Posters. find our card

↑ back to top