Elamur Research⌕
☰
home / papers / 2010 / A Tree Kernel-Based Unified Framework for Chinese Zero Anaphora Resolution

A Tree Kernel-Based Unified Framework for Chinese Zero Anaphora Resolution

Summary

Chinese is a pro-drop language, so subjects and objects are often omitted. These gaps (zero anaphors) carry no number or gender cues, and earlier computational work mostly assumed they were already located. Kong and Zhou cover the whole pipeline in one framework with three stages: finding zero anaphors, deciding whether each one has an antecedent in the text, and choosing that antecedent. Every stage uses a context-sensitive convolution tree kernel over parse tree structures built for that stage, which avoids hand-crafted features. To support the work, the authors annotated 100 documents from the Chinese Treebank (CTB) 6.0, labelling null constituents in the gold parses as anaphoric or non-anaphoric, and released the corpus. They reimplemented the feature-based system of Zhao and Ng (2007) as a baseline and evaluated it on the same data. The tree kernel models beat this baseline clearly on both anaphoricity determination and antecedent identification. This suggests that structural syntactic information is central to the task and that kernels capture it better than hand-built features.

Key points

  • Corpus: 100 CTB 6.0 documents (35,089 words) containing 1,720 zero anaphors, 712 of them anaphoric. Roughly 60% of zero anaphors have no antecedent, which makes anaphoricity determination essential.
  • Coverage: The authors describe this as the first systematic treatment of all three sub-tasks of Chinese zero anaphora resolution in one framework. Only zero anaphors licensed by verb phrase nodes are handled (about 93% of cases), and the experiments use manually parsed trees.
  • Baseline: Zhao and Ng (2007) was reproduced with Weka's J48 decision tree and its original feature sets.
  • Anaphoricity determination (gold zero anaphors): The baseline reaches 70.76 F, about 16 points below the tree kernel model. Precision shows the largest gap, at 63.61% versus 89.83%.
  • Antecedent identification: The baseline is about 15 F lower with gold anaphoric zero anaphors (62.20 F) and about 10 F lower in the fully automatic setting (35.43 F).

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

35 papers of our library cite it (of the 140673 whose bibliographies we read); 15 of them have a page here, 20 not yet.

Cite

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

@inproceedings{kong2010tree,
  title = {{A Tree Kernel-Based Unified Framework for Chinese Zero Anaphora Resolution}},
  author = {Kong, Fang and Zhou, Guodong},
  booktitle = {Proceedings of the 2010 Conference on Empirical Methods in Natural Language Processing},
  year = {2010},
  url = {https://aclanthology.org/D10-1086/}
}

References

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

  1. S. Converse. 2006. Pronominal Anaphora Resolution in Chinese. Ph.D. Thesis, Department of Computer and Information Science. University of Pennsylvania. find
  2. M. Collins and N. Duffy. 2001. Convolution kernels for natural language. NIPS’2001:625-632. find
  3. A. Ferrandez and J. Peral. 2000. A computational approach to zero-pronouns in Spanish. ACL'2000:166-172. find
  4. R. Iida, K. Inui, and Y. Matsumoto. 2006. Exploiting syntactic patterns as clues in zero-anaphora resolution. COLING-ACL'2006:625-632 find our card
  5. H. Isozaki and T. Hirao. 2003. Japanese zero pronoun resolution based on ranking rules and machine learning. EMNLP'2003:184-191 find our card
  6. F. Kong, G.D. Zhou and Q.M. Zhu. 2009 Employing the Centering Theory in Pronoun Resolution from the Semantic Perspective. EMNLP’2009: 987-996 find our card
  7. C. N. Li and S. A. Thompson. 1979. Third-person pronouns and zero-anaphora in Chinese discourse. Syntax and Semantics, 12:311-335. find
  8. W. Li. 2004. Topic chains in Chinese discourse. Discourse Processes, 37(1):25-45. find
  9. A. Moschitti. 2004. A Study on Convolution Kernels for Shallow Semantic Parsing, ACL’2004. find
  10. L.H. Qian, G.D. Zhou, F. Kong, Q.M. Zhu and P.D. Qian. 2008. Exploiting constituent dependencies for tree kernel-based semantic relation extraction. COLING’2008:697-704 find
  11. K. Seki, A. Fujii, and T. Ishikawa. 2002. A probabilistic method for analyzing Japanese anaphora intergrating zero pronoun detection and resolution. COLING'2002:911-917 find
  12. R. Sasano. D. Kawahara and S. Kurohashi. 2008. A fully-lexicalized probabilistic model for Japanese zero anaphora resolution. COLING'2008:769-776 find our card
  13. W.M. Soon, H.T. Ng and D. Lim. 2001. A machine learning approach to coreference resolution of noun phrase. find
  14. Computational Linguistics, 2001, 27(4):521-544. find
  15. V. Ng and C. Cardie 2002. Improving machine learning approaches to coreference resolution. ACL’2002: 104-111 find our card
  16. X.F. Yang, G.D. Zhou, J. Su and C.L. Chew. 2003. Coreference Resolution Using Competition Learning Approach. ACL’2003:177-184 find our card
  17. X.F. Yang, J. Su and C.L. Tan 2008. A Twin-Candidate Model for Learning-Based Anaphora Resolution. Computational Linguistics 34(3):327-356 find our card
  18. N. Xue, F. Xia, F.D. Chiou and M. Palmer. 2005. The Penn Chinese TreeBank: Phrase structure annotation of a large corpus. Natural Language Engineering, 11(2):207-238. find
  19. X.F. Yang, J. Su and C.L. Tan. 2006. Kernel-based pronoun resolution with structured syntactic knowledge. COLING-ACL'2006:41-48. find our card
  20. D. Zelenko, A. Chinatsu and R. Anthony. 2003. Kernel methods for relation extraction. Journal of Machine Learning Research, 3(2003):1083-1106 find
  21. S. Zhao and H.T. Ng. 2007. Identification and Resolution of Chinese Zero Pronouns: A Machine Learning Approach. EMNLP-CoNLL'2007:541-550. find our card
  22. S. Zhao and R. Grishman. 2005. Extracting relations with integrated information using kernel methods. ACL’2005:419-426 find
  23. G.D. Zhou, F. Kong and Q.M. Zhu. 2008. Context-sensitive convolution tree kernel for pronoun resolution. IJCNLP'2008:25-31 find our card
  24. G.D. Zhou, M. Zhang, D.H. Ji and Q.M. Zhu. 2007. Tree kernel-based relation extraction with context-sensitive structured parse tree information. EMNLP-CoNLL’2007:728-736 find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/kong-2010-tree-kernel-based-unified-framework/