Elamur Research⌕
☰
home / papers / 1999 / Noun Phrase Coreference as Clustering

Noun Phrase Coreference as Clustering

Summary

Cardie and Wagstaff address noun phrase coreference resolution, which means grouping the noun phrases in a text that refer to the same real-world entity. Most systems of the time relied on hand-written rules, and the few learning-based ones were supervised pairwise classifiers. The authors instead treat coreference as unsupervised clustering. Base noun phrases are found with the Empire NP finder, and each one becomes an 11-feature vector built with simple heuristics and WordNet. A weighted distance metric combines per-feature incompatibility scores:

  • Infinite weights act as hard filters.
  • Negative-infinite weights force a merge.
  • Weights equal to the clustering radius r express soft preferences that other evidence can override.

A greedy clustering procedure then links each NP to compatible earlier NPs that lie within r. On the MUC-6 coreference corpus, the method reaches an F-measure of 53.6%. This sits between the weakest and strongest MUC-6 systems and beats RESOLVE, the only learning-based participant. The result matters because it needs no annotated training data or full parse, and new linguistic knowledge can be added simply by extending the metric.

Key points

  • Results on MUC-6: the clustering approach scores F = 53.6%. For comparison, RESOLVE (decision trees) reaches 47%, the best MUC-6 system 65%, and the worst 40%.
  • Features: each base NP is described by 11 automatically derived features:
    • words and head noun
    • position
    • pronoun type and article
    • appositive status, number and proper-name status
    • semantic class and gender (both from WordNet)
    • animacy
  • Hard constraints: number, proper name, semantic class, gender and animacy work as hard filters. Appositive and words-substring terms force coreference.
  • Tuning and preprocessing: only the radius r and some weights are tuned on development data. The method needs neither a full syntactic parse nor a large statistics corpus.
  • Limitations and future work: errors come partly from greedy linking that can never be undone. The authors propose less greedy alternatives, such as linking only to the first or best-ranked antecedent, or processing links in ranked order. They also plan to learn the weights automatically and to use cues from centering theory.

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

42 papers of our library cite it (of the 140673 whose bibliographies we read); 16 of them have a page here, 26 not yet.

Cite

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

@inproceedings{cardie1999noun,
  title = {{Noun Phrase Coreference as Clustering}},
  author = {Cardie, Claire and Wagstaff, Kiri},
  booktitle = {1999 Joint SIGDAT Conference on Empirical Methods in Natural Language Processing and Very Large Corpora},
  year = {1999},
  url = {https://aclanthology.org/W99-0611/}
}

References

13 references, 5 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Chinatsu Aone and William Bennett. 1995. Evaluating Automated and Manual Acquisition of Anaphora Resolution Strategies. In Proceedings of the 33rd Annual Meeting of the ACL, pages 122-129. Association for Computational Linguistics. find our card
  2. S. Azzam, K. Humphreys, and R. Gaizauskas. 1998. Evaluating a Focus-Based Approach to Anaphora Resolution. In Proceedings of the 36th Annual Meeting of the ACL and COLING-98, pages 74-78. Association for Computational Linguistics. find our card
  3. C. Cardie and D. Pierce. 1998. Error-Driven Pruning of Treebank Grammars for Base Noun Phrase Identification. In Proceedings of the 36th Annual Meeting of the ACL and COLING-98, pages 218-224. Association for Computational Linguistics. find
  4. C. Cardie. 1992a. Corpus-Based Acquisition of Relative Pronoun Disambiguation Heuristics. In Proceedings off the 30th Annual Meeting off the ACL, pages 216-223, University of Delaware, Newark, DE. Association for Computational Linguistics. find
  5. C. Cardie. 1992b. Learning to Disambiguate Relative Pronouns. In! Proceedings of the Tenth National Conference on Artificial Intelligence, pages 38-43, San Jose, CA. AAAI Press / MIT Press. find
  6. I. Dagan and A. Itai. 1991. A Statistical Filter for Resolving Pron0un References. In Y. A. Feldman and A. Bruckstein, editors, Artificial Intelligence and Computer Vision, pages 125-135. Elsevier Science Publishers, North Holland. find
  7. C. Fellbaum. 1998. WordNet: An Electronical Lexical Database. The MIT Press, Cambridge, MA. find
  8. N. Ge, J. Hale, and E] Charniak. 1998. A Statistical Approach to Anaphora Resolution. In Charniak, Eugene, editor, Proceedings of the Sixth Workshop on Very Large Corpora, pages 161-170, Montreal, Canada. ACL SIGDAT. find our card
  9. S. Lappin and H. Leass. 1994. An Algorithm for Pronominal Anaphbra Resolution. Computational Linguistics, 20(4):5'35-562. find
  10. M. Marcus, M. Marcinkiewicz, and B. Santorini. 1993. Building a Large Annotated Corpus of English: The Penn Treebank. Computational Linguistics, 19(2):313-:330. find our card
  11. J. McCarthy and W. Lehnert. 1995. Using Decision Trees for Coreference Resolution. In C. Mellish, editor, Proceedings iof the Fourteenth International Conference on Art!ficial Intelligence, pages 1050-1055. 1995. Proceedings of the Sixth Message Understanding Conference (MUC-6). Morgan Kaufmann, San… find
  12. J. R. Quinlan. 1992. C~.5: Programs for Machine Learning. Morgan Kaufmaan, San Mateo, CA. find
  13. M. Vilaim J. Burger, J. Aberdeen, D. Connolly, and L. Hirschman. ~1995. A model-theoretic coreference scoring scheme. In Proceedings of the Sixth Message Understanding Conference (MUC-6), pages 45-52, San Francisco, CA. Morgan Kaufmann. find our card

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/cardie-1999-noun-phrase-coreference-as-clustering/