Elamur Research⌕
☰
home / papers / 2020 / Extending Implicit Discourse Relation Recognition to the PDTB-3

Extending Implicit Discourse Relation Recognition to the PDTB-3

Summary

The PDTB-3 adds many discourse relations, and most of the new ones fall inside single sentences. Some implicit relations are also now annotated alongside explicit connectives. Shallow discourse parsers built on the PDTB-2 assumed that non-explicit relations hold only between adjacent sentences, so their main difficulty was assigning senses. Liang, Zhao and Webber show that a PDTB-3 parser must also locate implicit relations. They argue that this extra work pays off, because intra-sentential, inter-sentential and explicit-linked implicit relations have distinct sense distributions that can guide sense classification. The authors document these distributional differences and build sense classifiers on an LSTM-based Basic Model, including variants specialised by relation location. They also add a simple recognizer that flags explicit relations carrying a co-occurring implicit relation. Their results indicate that knowing where an implicit relation sits improves sense labelling. The methods are not intended to be state-of-the-art. Instead, they provide a non-trivial baseline for future PDTB-3 shallow discourse parsers.

Key points

  • The PDTB-3 adds roughly 12.5K intra-sentential and about 1K inter-sentential relations compared with the PDTB-2. Over 95% of the new relations are sentence-internal, and 5780 of them are implicit.
  • The Basic Model encodes each argument with an LSTM followed by max-pooling, then passes the result through hidden, dense and softmax layers. It uses 300-dimension word2vec embeddings and a hidden size of 100.
  • Linked implicits are detected with an NLTK naive Bayes classifier. Its binary features are the 100 most frequent syntactic production rules. Recall for linked relations is low, since stand-alone relations dominate the data.
  • In the test set (WSJ Section 23), linked implicits mostly attach to conjoined clauses or VPs, and 81.08% are signalled by "and". All recognized linked implicits are intra-sentential.
  • On these recognized linked relations, the intra-sentential classifier reaches an F1 of 75, compared with 68.182 for the Basic Model.

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

12 papers of our library cite it (of the 140673 whose bibliographies we read); 8 of them have a page here, 4 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2020.codi-1.14 · arXiv 2010.06294

@inproceedings{liang2020extending,
  title = {{Extending Implicit Discourse Relation Recognition to the PDTB-3}},
  author = {Liang, Li and Zhao, Zheng and Webber, Bonnie},
  booktitle = {Proceedings of the First Workshop on Computational Approaches to Discourse},
  year = {2020},
  doi = {10.18653/v1/2020.codi-1.14},
  eprint = {2010.06294},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.18653/v1/2020.codi-1.14}
}

References

18 references, 15 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Nikita Kitaev and Dan Klein. 2018. Constituency parsing with a self-attentive encoder. In Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 2676–2686, Melbourne, Australia. Association for Computational Linguistics. ↗ our card
  2. Hongxiao Bai and Hai Zhao. 2018. Deep enhanced representation for implicit discourse relation recognition. In Proceedings of the 27th International Conference on Computational Linguistics, pages 571– 583, Santa Fe, New Mexico, USA. Association for Computational Linguistics. ↗ our card
  3. Christopher D Manning, Mihai Surdeanu, John Bauer, Jenny Rose Finkel, Steven Bethard, and David Mc-Closky. 2014. The stanford corenlp natural language processing toolkit. In Proceedings of 52nd annual meeting of the association for computational linguistics: system demonstrations, pages 55–60. find our card
  4. Mitchell P. Marcus, Beatrice Santorini, and Mary Ann Marcinkiewicz. 1993. Building a large annotated corpus of English: The Penn Treebank. Computational Linguistics, 19(2):313–330. ↗ our card
  5. Tomas Mikolov, Kai Chen, G.s Corrado, and Jeffrey Dean. 2013a. Efficient estimation of word representations in vector space. Proceedings of Workshop at ICLR, 2013. ↗
  6. Tomas Mikolov, Ilya Sutskever, Kai Chen, Greg Corrado, and Jeffrey Dean. 2013b. Distributed representations of words and phrases and their compositionality. In Proceedings of the 26th International Conference on Neural Information Processing Systems - Volume 2, NIPS’13, page 3111–3119, Red Hook, NY, USA. Curran Associates Inc. ↗
  7. Joonsuk Park and Claire Cardie. 2012. Improving implicit discourse relation recognition through feature set optimization. In Proceedings of the 13th Annual Meeting of the Special Interest Group on Discourse and Dialogue, pages 108–112, Seoul, South Korea. Association for Computational Linguistics. find our card
  8. Rashmi Prasad, Nikhil Dinesh, Alan Lee, Eleni Milt-sakaki, Livio Robaldo, Aravind Joshi, and Bonnie Webber. 2008. The penn discourse treebank 2.0. In Proceedings of the Sixth International Language Resources and Evaluation (LREC’08), pages 2961– 2968. European Language Resources Association (ELRA). find our card
  9. Rashmi Prasad, Aravind Joshi, and Bonnie Webber. 2010. Realization of discourse relations by other means: Alternative lexicalizations. In Proceedings of the 23rd International Conference on Computational Linguistics (COLING), Beijing, China. find our card
  10. Hannah Rohde, Anna Dickinson, Nathan Schneider, Christopher Clark, Annie Louis, and Bonnie Webber. 2017. Exploring substitutability through discourse adverbials and multiple judgments. In Proceedings, 12th International Conference on Computational Semantics (IWCS 2017), Montpellier, France. find
  11. Hannah Rohde, Alexander Johnson, Nathan Schneider, and Bonnie Webber. 2018. Discourse coherence: Concurrent explicit and implicit relations. In Proceedings of the 56 th Annual Meeting of the ACL. find our card
  12. Attapol Rutherford, Vera Demberg, and Nianwen Xue. 2017. A systematic study of neural discourse models for implicit discourse relation. In Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics (EACL 2017), pages 281–291. ↗ our card
  13. Attapol Rutherford and Nianwen Xue. 2016. Robust non-explicit neural discourse parser in English and Chinese. In Proceedings of the CoNLL-16 shared task, pages 55–59, Berlin, Germany. find our card
  14. Merel Scholman and Vera Demberg. 2017. Examples and specifications that prove a point: Identifying elaborative and argumentative discourse relations. Dialogue & Discourse, 8:56–83. ↗ our card
  15. Wei Shi and Vera Demberg. 2017. On the need of cross validation for discourse relation classification. In Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 2, Short Papers, pages 150–156, Valencia, Spain. Association for Computational Linguistics. ↗ our card
  16. Bonnie Webber, Rashmi Prasad, Alan Lee, and Aravind Joshi. 2019. The penn discourse treebank 3.0 annotation manual. https://catalog.ldc.upenn.edu/docs/LDC2019T05/ PDTB3-Annotation-Manual.pdf . ↗ our card
  17. Nianwen Xue, Hwee Tou Ng, Sameer Pradhan, Rashmi Prasad, Christopher Bryant, and Attapol Rutherford. 2015. The CoNLL-2015 shared task on shallow discourse parsing. In Proceedings of the Nineteenth Conference on Computational Natural Language Learning - Shared Task, pages 1–16, Beijing, China. Association for Computational Linguistics. find our card
  18. Nianwen Xue, Hwee Tou Ng, Sameer Pradhan, Attapol Rutherford, Bonnie Webber, Chuan Wang, and Hongmin Wang. 2016. CoNLL 2016 shared task on multilingual shallow discourse parsing. In Proceedings of the CoNLL-16 shared task, pages 1– 19, Berlin, Germany. Association for Computational Linguistics. ↗ our card

↑ back to top

This page in Markdown →