Elamur Research⌕
☰
home / papers / 2011 / A Discriminative Approach to Japanese Zero Anaphora Resolution with Large-scale Lexicalized Case Frames

A Discriminative Approach to Japanese Zero Anaphora Resolution with Large-scale Lexicalized Case Frames

Summary

Japanese often omits predicate arguments, so finding the referents of these "zero pronouns" is essential for recovering predicate-argument structure. Sasano and Kurohashi target Web text, which is more stylistically varied than the newspaper data used in most earlier work. They build on the probabilistic model of Sasano et al. (2008), whose independence assumptions made it hard to add overlapping features or weight their importance. Their replacement is a log-linear model that jointly selects a case frame for a predicate and its full predicate-argument structure, covering both overt and omitted arguments.

The model draws on two kinds of evidence:

  • Lexical knowledge from case frames built automatically from about 1.6 billion Web sentences. Slot fillers are generalized through noun categories, NE classes and automatically induced noun clusters.
  • Non-lexical cues, such as case/location and salience, learned from a small annotated corpus.

On the Web Corpus the model outperformed the earlier approach. It also performed comparably to prior work on newspaper text. Feature analysis shows that lexical preferences differ across cases.

Key points

  • Case frames. The frames were built with Kawahara and Kurohashi (2002)'s method, with separate frames for each sense of a predicate. Slot examples are generalized using:
    • 22 JUMAN noun categories,
    • 8 IREX NE classes,
    • 2,000 clusters (Kazama and Torisawa, 2008) derived from 1 million multi-word nouns.
  • Scope. Resolution is limited to three cases: nominative (ga), accusative (wo) and dative (ni).
  • Main result. On the 20 test articles used by Sasano et al. (2008), the model reached an F-measure of 0.566 (R 0.525, P 0.615). The earlier model scored 0.391.
  • Ablation. Removing case/location features lowered F by 0.092, and removing word PMI lowered it by 0.021. Generalized PMIs (cluster, category, NE) help collectively, and salience features became redundant once case/location features were included.
  • Newspaper comparison. On the NAIST Text Corpus 1.4β, results were comparable to Imamura et al. (2009). Performance seemed lower than Iida et al. (2006)'s syntactic-pattern approach, though the evaluation data differed.

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 · Semantic role labeling

Cited by

33 papers of our library cite it (of the 140673 whose bibliographies we read); 13 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{sasano2011discriminative,
  title = {{A Discriminative Approach to Japanese Zero Anaphora Resolution with Large-scale Lexicalized Case Frames}},
  author = {Sasano, Ryohei and Kurohashi, Sadao},
  booktitle = {Proceedings of 5th International Joint Conference on Natural Language Processing},
  year = {2011},
  url = {https://aclanthology.org/I11-1085/}
}

References

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

  1. A. Ferrandez and J. Peral. 2000. A computational approach to zero-pronouns in spanish. In Proc. of ACL’00, pages 166–172. find
  2. C. J. Fillmore, C. R. Johnson, and M. R.L. Petruck. 2003. Background to framenet. International Journal of Lexicography, 16(3):235–250. find
  3. Aria Haghighi and Dan Klein. 2010. Coreference resolution in a modular, entity-centered model. In Proc. of NAACL-HLT’10, pages 385–393. find our card
  4. Ryu Iida, Kentaro Inui, and Yuji Matsumoto. 2006. Exploiting syntactic patterns as clues in zero-anaphora resolution. In Proc. of COLING-ACL’06, pages 625–632. find our card
  5. Ryu Iida, Mamoru Komachi, Kentaro Inui, and Yuji Matsumoto. 2007. Annotating a japanese text corpus with predicate-argument and coreference relations. In Proc. of ACL’07 Workshop: Linguistic Annotation Workshop, pages 132–139. find our card
  6. Kenji Imamura, Kuniko Saito, and Tomoko Izumi. 2009. Discriminative approach to predicate-argument structure analysis with zero-anaphora resolution. In Proc. of ACL-IJCNLP’09, pages 85–88. find our card
  7. IREX Committee, editor. 1999. Proc. of the IREX Workshop. find
  8. Daisuke Kawahara and Sadao Kurohashi. 2002. Fertilization of case frame dictionary for robust Japanese case analysis. In Proc. of COLING’02, pages 425– 431. find
  9. Daisuke Kawahara and Sadao Kurohashi. 2004. Zero pronoun resolution based on automatically constructed case frames and structural preference of antecedents. In Proc. of IJCNLP’04, pages 334–341. find
  10. Daisuke Kawahara and Sadao Kurohashi. 2006. A fully-lexicalized probabilistic model for Japanese syntactic and case structure analysis. In Proc. of HLT-NAACL’06, pages 176–183. find
  11. Jun’ichi Kazama and Kentaro Torisawa. 2008. Inducing gazetteers for named entity recognition by large-scale clustering of dependency relations. In Proc. of ACL-HLT’08, pages 407–415. find
  12. Fang Kong and Guodong Zhou. 2010. A tree kernel-based unified framework for chinese zero anaphora resolution. In Proc. of EMNLP’10, pages 882–891. find our card
  13. Anna Korhonen, Yuval Krymolowski, and Ted Briscoe. 2006. A large subcategorization lexicon for natural language processing applications. In Proc.of LREC’06, pages 3000–3006. find
  14. Shalom Lappin and Herbert J. Leass. 1994. An algorithm for pronominal anaphora resolution. Computational Linguistics, 20(4):535–562. find our card
  15. Ruslan Mitkov, Richard Evans, and Constantin Orăsan. 2002. A new, fully automatic version of Mitkov’s knowledge-poor pronoun resolution method. In Proc. of CICLing’02, pages 168–186. find
  16. Vincent Ng and Claire Cardie. 2002. Improving machine learning approaches to coreference resolution. In Proc. of ACL’02, pages 104–111. find our card
  17. Jorge Nocedal. 1980. Updating quasi-newton matrices with limited storage. Mathematics of Computation, 35(151):773–782. find
  18. Martha Palmer, Dan Gildea, and Paul Kingsbury. 2005. The proposition bank: A corpus annotated with semantic roles. Computational Linguistics, 31(1):71– 105. find
  19. Ryohei Sasano, Daisuke Kawahara, and Sadao Kurohashi. 2008. A fully-lexicalized probabilistic model for japanese zero anaphora resolution. In Proc. of COLING’08, pages 769–776. find our card
  20. Ryohei Sasano, Daisuke Kawahara, and Sadao Kurohashi. 2009. The effect of corpus size on case frame acquisition for discourse analysis. In Proc. of NAACL-HLT’09, pages 521–529. find
  21. Kazuhiro Seki, Atsushi Fujii, and Tetsuya Ishikawa. 2002. A probabilistic method for analyzing Japanese anaphora integrating zero pronoun detection and resolution. In Proc. of COLING’02, pages 911–917. find our card
  22. Hirotoshi Taira, Sanae Fujita, and Masaaki Nagata. 2008. A japanese predicate argument structure analysis using decision lists. In Proc. of EMNLP’08, pages 523–532. find
  23. Shanheng Zhao and Hwee Tou Ng. 2007. Identification and resolution of Chinese zero pronouns: A machine learning approach. In Proc. of EMNLP-CoNLL’07, pages 541–550. find our card

↑ back to top