Elamur Research⌕
☰
home / papers / 2007 / Semantic Class Induction and Coreference Resolution

Semantic Class Induction and Coreference Resolution

Summary

This paper asks whether a learning-based NP coreference resolver performs better when it uses semantic class (SC) knowledge learned from annotated data instead of the common shortcut of taking a noun's first WordNet sense. Ng trains a six-way SC classifier (PERSON, ORGANIZATION, GPE, FACILITY, LOCATION, OTHERS) on the BBN Entity Type Corpus, a version of the Penn Treebank WSJ texts in which ACE mentions carry SC labels. The features combine lexical cues, MINIPAR grammatical relations, IdentiFinder NE types, WordNet hypernym keywords, classes induced from appositive constructions in large unlabeled corpora, and neighbors from Lin's distributional thesaurus.

The predicted classes feed two knowledge sources: semantic class agreement (SCA) and mention status. Each is added to a decision-tree baseline either as a hard constraint or as a feature. On ACE Phase 2, the induced classes beat Soon et al.'s first-sense heuristic in both SC accuracy and coreference F-measure. The study also shows that how knowledge is encoded matters as much as which knowledge is used.

Key points

  • Training data: The SC classifier learns from the BBN Entity Type Corpus, which offers roughly five times more training material than ACE and yields 310,063 instances. Four learners were tested, among them Decision List, 1-NN (TiMBL) and SVM.
  • Gains over the heuristic: Induced SCs improve SC accuracy by about 2% over the first-sense heuristic. They also give a statistically significant 2% F-measure gain in coreference over a resolver using Soon et al.'s SCs.
  • Mention as a constraint: Filtering out non-mentions raises MUC F-measure by roughly 5–7%. For example, the baseline scores 57.0, Soon's SCs reach 62.6, and DL or SVM SCs reach 64.6.
  • SCA as a feature: Encoding SCA as a feature lifts common noun resolution accuracy by about 5–8% over the baseline. The strongest overall configuration applies mention status as a hard constraint while treating SCA as a feature, and this remains true even with oracle SCs.
  • Evaluation lesson: The MUC scorer fails to reveal the benefit of SCA. The author therefore argues for reporting both resolution accuracy and clustering-level scores.

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

21 papers of our library cite it (of the 140673 whose bibliographies we read); 10 of them have a page here, 11 not yet.

Cite

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

@inproceedings{ng2007semantic,
  title = {{Semantic Class Induction and Coreference Resolution}},
  author = {Ng, Vincent},
  booktitle = {Proceedings of the 45th Annual Meeting of the Association of Computational Linguistics},
  year = {2007},
  url = {https://aclanthology.org/P07-1068/}
}

References

28 references, 13 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. D. Bean and E. Riloff. 2004. Unsupervised learning of contextual role knowledge for coreference resolution. In Proc. of HLT/NAACL, pages 297–304. find our card
  2. D. M. Bikel, R. Schwartz, and R. M. Weischedel. 1999. An algorithm that learns what’s in a name. Machine Learning 34(1–3):211–231. find
  3. C.-C. Chang and C.-J. Lin, 2001. LIBSVM: a library for support vector machines. Software available at http://www.csie.ntu.edu.tw/∼cjlin/libsvm. ↗
  4. M. Collins and Y. Singer. 1999. Unsupervised models for named entity classification. In Proc. of EMNLP/VLC. find
  5. W. Daelemans, J. Zavrel, K. van der Sloot, and A. van den Bosch. 2004. TiMBL: Tilburg Memory Based Learner, version 5.1, Reference Guide. ILK Technical Report. find
  6. H. Daumé III and D. Marcu. 2005. A large-scale exploration of effective global features for a joint entity detection and tracking model. In Proc. of HLT/EMNLP, pages 97–104. find
  7. R. Florian, H. Jing, N. Kambhatla, and I. Zitouni. 2006. Factorizing complex models: A case study in mention detection. In Proc. of COLING/ACL, pages 473–480. find
  8. M. Hearst. 1992. Automatic acquisition of hyponyms from large text corpora. In Proc. of COLING. find
  9. H. Ji, D. Westbrook, and R. Grishman. 2005. Using semantic relations to refine coreference decisions. In Proc. of HLT/EMNLP, pages 17–24. find our card
  10. A. Kehler, D. Appelt, L. Taylor, and A. Simma. 2004. The (non)utility of predicate-argument frequencies for pronoun interpretation. In Proc. of NAACL, pages 289–296. find
  11. D. Lin. 1998a. Automatic retrieval and clustering of similar words. In Proc. of COLING/ACL, pages 768–774. find
  12. D. Lin. 1998b. Dependency-based evaluation of MINIPAR. In Proc. of the LREC Workshop on the Evaluation of Parsing Systems, pages 48–56. find
  13. D. Lin. 1998c. Using collocation statistics in information extraction. In Proc. of MUC-7. find
  14. X. Luo, A. Ittycheriah, H. Jing, N. Kambhatla, and S. Roukos. 2004. A mention-synchronous coreference resolution algorithm based on the Bell tree. In Proc. of the ACL. find our card
  15. K. Markert and M. Nissim. 2005. Comparing knowledge sources for nominal anaphora resolution. Computational Linguistics, 31(3):367–402. find our card
  16. R. Mitkov. 2002. Anaphora Resolution. Longman. find
  17. R. Mitkov. 1998. Robust pronoun resolution with limited knowledge. In Proc. of COLING/ACL, pages 869–875. find our card
  18. V. Ng and C. Cardie. 2002. Improving machine learning approaches to coreference resolution. In Proc. of the ACL. find our card
  19. E. W. Noreen. 1989. Computer Intensive Methods for Testing Hypothesis: An Introduction. John Wiley & Sons. find
  20. M. Poesio, R. Mehta, A. Maroudas, and J. Hitzeman. 2004. Learning to resolve bridging references. In Proc. of the ACL. find our card
  21. S. P. Ponzetto and M. Strube. 2006. Exploiting semantic role labeling, WordNet and Wikipedia for coreference resolution. In Proc. of HLT/NAACL, pages 192–199. find our card
  22. J. R. Quinlan. 1993. C4.5: Programs for Machine Learning. Morgan Kaufmann, San Mateo, CA. find our card
  23. W. M. Soon, H. T. Ng, and D. Lim. 2001. A machine learning approach to coreference resolution of noun phrases. Computational Linguistics, 27(4):521–544. find our card
  24. J. Tetreault. 2001. A corpus-based evaluation of centering and pronoun resolution. Computational Linguistics, 27(4). find our card
  25. M. Vilain, J. Burger, J. Aberdeen, D. Connolly, and L. Hirschman. 1995. A model-theoretic coreference scoring scheme. In Proc. of MUC-6, pages 45–52. find our card
  26. R. Weischedel and A. Brunstein. 2005. BBN pronoun coreference and entity type corpus. Linguistica Data Consortium. find
  27. X. Yang, G. Zhou, J. Su, and C. L. Tan. 2003. Coreference resolution using competitive learning approach. In Proc. of the ACL, pages 176–183. find
  28. D. Yarowsky. 1995. Unsupervised word sense disambiguation rivaling supervised methods. In Proc. of the ACL. find our card

↑ back to top

This page in Markdown →