Elamur Research⌕
☰
home / papers / 2014 / Discriminative Reranking of Discourse Parses Using Tree Kernels

Discriminative Reranking of Discourse Parses Using Tree Kernels

Summary

The paper tackles RST-style discourse parsing. Existing parsers, including the then state-of-the-art system of Joty et al. (2013), mostly rely on local decisions and miss long-range structural dependencies between discourse tree (DT) constituents. The authors modify that probabilistic parser so it outputs k-best DT hypotheses. They then rerank these hypotheses with Support Vector Machines in a preference-ranking setup. The key ingredient is a family of discourse tree kernels (DiscTK), built by combining standard tree kernels (STK, PTK) with new tree encodings of DTs (JRN and SRN). This lets the reranker implicitly use every subtree fragment as a feature, with no manual feature design. On RST-DT, sentence-level parsing f-score rises from 79.77% to 82.15%. Plugging this reranker into the document-level parser raises its accuracy from 55.8% to 57.3%, a new state of the art. The study shows that structural features matter for discourse parsing and that tree kernels offer a practical way to capture them.

Key points

  • Data and setup: experiments use the RST Discourse Treebank (RST-DT) with manual EDU segmentation, on the standard test split plus 5-fold cross-validation.
  • Sentence-level gains: the best reranker (k=4–5) reaches 82.15% f-score, a relative error reduction of about 11.8%. The 5-best oracle is 88.09%, and human agreement is around 83%.
  • Document-level gains: accuracy improves from 55.8% to 57.3% (p < 0.01). However, k-best oracles at this level are low (56.91% for 5-best, 57.65% for 20-best), which limits how much reranking can help.
  • Representations and kernels: SRN outperforms JRN, and STK works better than PTK. Feature-vector rerankers using basic features or bag-of-relations bring no real gain, whereas tree fragments do.
  • Error analysis: the reranker struggles with small DTs of two or three EDUs whose relations are semantically close (e.g., Summary vs. Elaboration, Cause vs. Elaboration). The authors conclude that richer semantic representations are needed.

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

10 papers of our library cite it (of the 140673 whose bibliographies we read); 3 of them have a page here, 7 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.3115/v1/d14-1219

@inproceedings{joty2014discriminative,
  title = {{Discriminative Reranking of Discourse Parses Using Tree Kernels}},
  author = {Joty, Shafiq and Moschitti, Alessandro},
  booktitle = {Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP)},
  year = {2014},
  doi = {10.3115/v1/d14-1219},
  url = {https://doi.org/10.3115/v1/d14-1219}
}

References

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

  1. Lynn Carlson and Daniel Marcu. 2001. Discourse Tagging Reference Manual. Technical Report ISI-TR-545, University of Southern California Information Sciences Institute. find
  2. Lynn Carlson, Daniel Marcu, and Mary Ellen Okurowski. 2002. RST Discourse Treebank (RST-DT) LDC2002T07. Linguistic Data Consortium, Philadelphia. find
  3. Eugene Charniak and Mark Johnson. 2005. Coarse-to-Fine n-Best Parsing and MaxEnt Discriminative Reranking. In Proceedings of the 43rd Annual Meeting of the Association for Computational Linguistics, ACL’05, pages 173–180, NJ, USA. ACL. find our card
  4. Michael Collins and Nigel Duffy. 2002. New Ranking Algorithms for Parsing and Tagging: Kernels over Discrete Structures, and the Voted Perceptron. In ACL. find
  5. Michael Collins and Terry Koo. 2005. Discriminative Reranking for Natural Language Parsing. Comput. Linguist., 31(1):25–70, March. find our card
  6. Corinna Cortes and Vladimir Vapnik. 1995. Support Vector Networks. Machine Learning, 20:273–297. find
  7. Marco Dinarelli, Alessandro Moschitti, and Giuseppe Riccardi. 2011. Discriminative Reranking for Spoken Language Understanding. IEEE Transactions on Audio, Speech and Language Processing (TASLP), 20:526539. find
  8. Vanessa Feng and Graeme Hirst. 2012. Text-level Discourse Parsing with Rich Linguistic Features. In Proceedings of the 50th Annual Meeting of the Association for Computational Linguistics, ACL ’12, pages 60–68, Jeju Island, Korea. ACL. find our card
  9. Hugo Hernault, Helmut Prendinger, David A. duVerle, and Mitsuru Ishizuka. 2010. HILDA: A Discourse Parser Using Support Vector Machine Classification. Dialogue and Discourse, 1(3):1—33. find our card
  10. Liang Huang and David Chiang. 2005. Better Kbest Parsing. In Proceedings of the Ninth International Workshop on Parsing Technology, Parsing ’05, pages 53–64, Stroudsburg, PA, USA. Association for Computational Linguistics. find
  11. Thorsten Joachims and Chun-Nam John Yu. 2009. Sparse Kernel SVMs via Cutting-Plane Training. Machine Learning, 76(2-3):179–193. ECML. find
  12. Thorsten Joachims. 1999. Making large-Scale SVM Learning Practical. In Advances in Kernel Methods - Support Vector Learning. find
  13. Shafiq Joty, Giuseppe Carenini, and Raymond T. Ng. 2012. A Novel Discriminative Framework for Sentence-Level Discourse Analysis. In Proceedings of the 2012 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning, EMNLP-CoNLL ’12, pages 904–915… find
  14. Shafiq Joty, Giuseppe Carenini, Raymond T. Ng, and Yashar Mehdad. 2013. Combining Intra- and Multi-sentential Rhetorical Parsing for Document-level Discourse Analysis. In Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics, ACL ’13, Sofia, Bulgaria. ACL. find
  15. William Mann and Sandra Thompson. 1988. Rhetorical Structure Theory: Toward a Functional Theory of Text Organization. Text, 8(3):243–281. find our card
  16. Daniel Marcu. 2000a. The Rhetorical Parsing of Unrestricted Texts: A Surface-based Approach. Computational Linguistics, 26:395–448. find our card
  17. Daniel Marcu. 2000b. The Theory and Practice of Discourse Parsing and Summarization. MIT Press, Cambridge, MA, USA. find
  18. Alessandro Moschitti, Daniele Pighin, and Roberto Basili. 2006. Semantic Role Labeling via Tree Kernel Joint Inference. In Proceedings of the Tenth Conference on Computational Natural Language Learning (CoNLL-X), pages 61–68, New York City, June. Association for Computational Linguistics. find
  19. Alessandro Moschitti. 2006. Efficient Convolution Kernels for Dependency and Constituent Syntactic Trees. In 17th European Conference on Machine Learning, pages 318–329. Springer. find
  20. Alessandro Moschitti. 2010. Kernel Engineering for Fast and Easy Design of Natural Language Applications. In COLING (Tutorials), pages 1–91. find
  21. Alessandro Moschitti. 2012. State-of-the-Art Kernels for Natural Language Processing. In Tutorial Abstracts of ACL 2012, page 2, Jeju Island, Korea, July. Association for Computational Linguistics. find
  22. Alessandro Moschitti. 2013. Kernel-based Learning to Rank with Syntactic and Semantic Structures. In SIGIR, page 1128. find
  23. Daniele Pighin and Alessandro Moschitti. 2009. Reverse Engineering of Tree Kernel Feature Spaces. In EMNLP, pages 111–120. find
  24. Daniele Pighin and Alessandro Moschitti. 2010. On Reverse Feature Engineering of Syntactic Tree Kernels. In Proceedings of the Fourteenth Conference on Computational Natural Language Learning, pages 223–233, Uppsala, Sweden, July. Association for Computational Linguistics. find
  25. Aliaksei Severyn and Alessandro Moschitti. 2011. Fast Support Vector Machines for Structural Kernels. In ECML/PKDD (3), pages 175–190. find
  26. Aliaksei Severyn and Alessandro Moschitti. 2012. Fast Support Vector Machines for Convolution Tree Kernels. Data Min. Knowl. Discov., 25(2):325–357. find
  27. John Shawe-Taylor and Nello Cristianini. 2004. Kernel Methods for Pattern Analysis. Cambridge University Press. find
  28. Radu Soricut and Daniel Marcu. 2003. Sentence Level Discourse Parsing Using Syntactic and Lexical Information. In Proceedings of the 2003 Conference of the North American Chapter of the Association for Computational Linguistics on Human Language Technology - Volume 1, NAACL’03, pages 149–156, Edmont… find our card
  29. Caroline Sporleder and Mirella Lapata. 2005. Discourse Chunking and its Application to Sentence Compression. In Proceedings of the conference on Human Language Technology and Empirical Methods in Natural Language Processing, HLT-EMNLP’05, pages 257–264, Vancouver, British Columbia, Canada. ACL. find
  30. Charles Sutton, Andrew McCallum, and Khashayar Rohanimanesh. 2007. Dynamic Conditional Random Fields: Factorized Probabilistic Models for Labeling and Segmenting Sequence Data. Journal of Machine Learning Research (JMLR), 8:693–723. find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/papers/joty-2014-discriminative-reranking-discourse-parses-tree/