Elamur Research⌕
☰
home / papers / 2008 / A Fully-Lexicalized Probabilistic Model for Japanese Zero Anaphora Resolution

A Fully-Lexicalized Probabilistic Model for Japanese Zero Anaphora Resolution

Summary

Japanese frequently omits arguments of predicates, leaving "zero pronouns" that must be both detected and resolved to an antecedent. The authors argue these two steps are interdependent and propose a single probabilistic model that performs case structure analysis and zero anaphora resolution jointly. The model relies on wide-coverage case frames built automatically from roughly 1.6 billion web sentences. To reduce sparseness, case-slot examples are generalized using JUMAN common-noun categories and IREX named-entity classes. Unlike most earlier systems, zero pronouns are linked to discourse entities rather than to individual mentions. This design makes entity salience usable as a filter on antecedent candidates. The model was evaluated on a newly annotated corpus of Japanese web documents, using gold preprocessing. It reached an F-measure of about 0.39, roughly double the earlier Kawahara & Kurohashi (2004) approach on the same data. The work shows that fully lexicalized, frequency-aware case frames can handle zero anaphora in noisy web text, a genre harder than newspapers.

Key points

  • Case frames: built from about 1.6 billion web sentences, following Kawahara and Kurohashi (2002). Slot examples are enriched with rates for about twenty JUMAN categories (e.g., [CT:PERSON]) and eight IREX NE classes (e.g., [NE:ORGANIZATION]).
  • Scope: the model covers the ga, wo and ni case slots, which account for about 90% of zero anaphora. It uses KNP parsing and gold morphology, NEs, syntax and coreference.
  • Data: a tagged corpus of 186 web documents (979 sentences). 166 documents were used for estimating probabilities and 20 for testing, with 122 zero anaphora relations in the test set.
  • Results: the best setting (α = 1/2) achieved R .410, P .373 and F .391, compared with F .197 for Kawahara & Kurohashi (2004). Most errors came from undetected zero pronouns: 60 of the 72 misses were never detected.
  • Ablation: generalized categories gave the largest gain, while NE generalization helped only slightly. The salience filter raised F from .260 to .391 and made processing about ten times faster.

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

26 papers of our library cite it (of the 140673 whose bibliographies we read); 13 of them have a page here, 13 not yet.

Cite

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

@inproceedings{sasano2008fully,
  title = {{A Fully-Lexicalized Probabilistic Model for Japanese Zero Anaphora Resolution}},
  author = {Sasano, Ryohei and Kawahara, Daisuke and Kurohashi, Sadao},
  booktitle = {Proceedings of the 22nd International Conference on Computational Linguistics (Coling 2008)},
  year = {2008},
  url = {https://aclanthology.org/C08-1097/}
}

References

15 references, 8 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Iida, Ryu, Kentaro Inui, and Yuji Matsumoto. 2006. Exploiting syntactic patterns as clues in zero-anaphora resolution. In Proceedings of COL-ING/ACL 2006, pages 625–632. find our card
  2. IREX Committee, editor. 1999. Proc. of the IREX Workshop. find
  3. Isozaki, Hideki and Tsutomu Hirao. 2003. Japanese zero pronoun resolution based on ranking rules and machine learning. In Proceedings of the Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 184–191. find our card
  4. Kawahara, Daisuke and Sadao Kurohashi. 2002. Fertilization of Case Frame Dictionary for Robust Japanese Case Analysis. In Proceedings of the 19th International Conference on Computational Linguistics, pages 425–431. find
  5. Kawahara, Daisuke and Sadao Kurohashi. 2004. Zero pronoun resolution based on automatically constructed case frames and structural preference of antecedents. In Proceedings of the 1st International Joint Conference on Natural Language Processing (IJCNLP-04), pages 334–341. find
  6. Kawahara, Daisuke and Sadao Kurohashi. 2006. A fully-lexicalized probabilistic model for japanese syntactic and case structure analysis. In Proceedings of the Human Language Technology Conference of the NAACL, Main Conference, pages 176–183. find
  7. Lappin, Shalom and Herbert J. Leass. 1994. An algorithm for pronominal anaphora resolution. Computational Linguistics, 20(4):535–562. find our card
  8. Luo, Xiaoqiang. 2007. Coreference or not: A twin model for coreference resolution. In Human Language Technologies 2007: The Conference of the North American Chapter of the Association for Computational Linguistics; Proceedings of the Main Conference, pages 73–80. find our card
  9. Mitkov, Ruslan, Richard Evans, and Constantin Orăsan. 2002. A new, fully automatic version of mitkov’s knowledge-poor pronoun resolution method. In Proceedings of the Third International Conference on Intelligent Text Processing and Computational Linguistics (CICLing-2002). find
  10. Ng, Vincent. 2005. Machine learning for coreference resolution: From local classification to global ranking. In Proceedings of the 43rd Annual Meeting of the Asssociation for Computational Linguistics, pages 157–164. find our card
  11. Sasano, Ryohei and Sadao Kurohashi. 2008. Japanese named entity recognition using structural natural language processing. In Proceedings of the 3rd International Joint Conference on Natural Language Processing (IJCNLP-08), pages 607–612. find
  12. Seki, Kazuhiro, Atsushi Fujii, and Tetsuya Ishikawa. 2002. A probabilistic method for analyzing Japanese anaphora integrating zero pronoun detection and resolution. In Proceedings of the 19th International Conference on Computational Linguistics, pages 911–917. find our card
  13. Soon, Wee Meng, Hwee Tou Ng, and Daniel Chung Yong Lim. 2001. A machine learning approach to coreference resolution of noun phrases. Computational Linguistics, 27(4):521–544. find our card
  14. The National Language Institute for Japanese Language. 2004. Bunruigoihyo. Dainippon Tosho, (In Japanese). find
  15. Yang, Xiaofeng, Jian Su, Jun Lang, Ghew Lim Tan, Ting Liu, and Sheng Li. 2008. An entity-mention model for coreference resolution with inductive logic programming. In Proceedings of ACL-08: HLT, pages 843–851. find our card

↑ back to top