Elamur Research⌕
☰
home / papers / 2003 / Using the Web in Machine Learning for Other-Anaphora Resolution

Using the Web in Machine Learning for Other-Anaphora Resolution

Summary

The paper addresses other-anaphors: noun phrases modified by "other" or "another" whose antecedents must be found in the preceding discourse rather than in the same syntactic structure. The authors built a corpus of 500 such anaphors from the Wall Street Journal portion of the Penn Treebank (release 2), with manually annotated NP antecedents. They then trained a Naive Bayes classifier from Weka on anaphor–candidate pairs.

The first feature set (F1) has nine features covering surface form, grammatical role, sentence distance and string matching. It also includes semantic information drawn from WordNet and the ANNIE named-entity tagger in GATE2. Error analysis showed that this lexical knowledge often falls short, because of rare word senses, missing relations and context-dependent links.

To close this gap, the authors queried Google with lexico-syntactic patterns specific to other-anaphora. They turned the page counts into a mutual-information score and marked the best-scoring candidate for each anaphor. Adding this cheap Web feature clearly improved results. The work is the first machine-learning approach to anaphora resolution that draws on the Web, and it can be adapted to other anaphoric phenomena.

Key points

  • Data: 500 other-anaphors from the Wall Street Journal (Penn Treebank), giving 3,084 anaphor–antecedent pairs, of which 16% are positive. Evaluation used 10-fold cross-validation.
  • A string-matching and recency baseline reached 27.8% P/R/F. Naive Bayes with F1 reached P 51.7, R 40.6 and F 45.5.
  • With the Web feature (F1+Web), scores rose to P 60.8, R 53.4 and F 56.9. This is a gain of 11.4 F-points, about 25% relative.
  • The Web feature labels the top-scoring candidate "webfirst" and all other candidates "webrest", which introduces competition among antecedents.
  • Remaining errors come from patterns too rare even on the Web and from mistakes in named-entity classification (39.6% of antecedents are named entities). Pronoun candidates are not yet handled.

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: Named entity recognition · Coreference resolution

Cited by

17 papers of our library cite it (of the 140673 whose bibliographies we read); 10 of them have a page here, 7 not yet.

Cite

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

@inproceedings{modjeska2003using,
  title = {{Using the Web in Machine Learning for Other-Anaphora Resolution}},
  author = {Modjeska, Natalia N. and Markert, Katja and Nissim, Malvina},
  booktitle = {Proceedings of the 2003 Conference on Empirical Methods in Natural Language Processing},
  year = {2003},
  doi = {10.3115/1119355.1119378},
  url = {https://doi.org/10.3115/1119355.1119378}
}

References

19 references, 4 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. C. Aone and S. W. Bennett. 1995. Evaluating automated and manual acquisition of anaphora resolution strategies. In Proc. of ACL’95, pages 122–129. find our card
  2. M. Berland and E. Charniak. 1999. Finding parts in very large corpora. In Proc. of ACL’99, pages 57–64. find
  3. G. Bierner. 2001. Alternative phrases and natural language information retrieval. In Proc. of ACL’01. find
  4. R. Bunescu. 2003. Associative anaphora resolution: A Web-based approach. In R. Dale, K. van Deemter, and R. Mitkov, editors, Proc. of the EACL Workshop on the Computational Treatment of Anaphora. find
  5. D. Connolly, J. D. Burger, and D. S. Day. 1997. A machine learning approach to anaphoric reference. In Daniel Jones and Harold Somers, editors, New Methods in Language Processing, pages 133–144. UCL Press, London. find
  6. C. Fellbaum, editor. 1998. WordNet: An Electronic Lexical Database. The MIT Press. find
  7. G. Grefenstette. 1999. The WWW as a resource for example-based MT tasks. In Proc. of ASLIB’99 Translating and the Computer 21, London. find
  8. S. Harabagiu, G. Miller, and D. Moldovan. 1999. Wordnet 2 - a morphologically and semantically enhanced resource. In Proc. of SIGLEX-99, pages 1–8. find
  9. M. Hearst. 1992. Automatic acquisition of hyponyms from large text corpora. In Proc. of COLING-92. find
  10. F. Keller, M. Lapata, and O. Ourioupina. 2002. Using the Web to overcome data sparseness. In Proc. of EMNLP 2002, pages 230–237. find
  11. K. Markert, M. Nissim, and N. N. Modjeska. 2003. Using the Web for nominal anaphora resolution. In R. Dale, K. van Deemter, and R. Mitkov, editors, Proc. of the EACL Workshop on the Computational Treatment of Anaphora, pages 39–46. find
  12. J. F. McCarthy and W. G. Lehnert. 1995. Using decision trees for coreference resolution. In Proc. of IJCAI-95, pages 1050–1055. find
  13. N. N. Modjeska. 2002. Lexical and grammatical role constraints in resolving other-anaphora. In Proc. of DAARC 2002, pages 129–134. find
  14. V. Ng and C. Cardie. 2002. Improving machine learning approaches to coreference resolution. In Proc. of ACL’02, pages 104–111. find our card
  15. M. Poesio, T. Ishikawa, S. Schulte im Walde, and R. Viera. 2002. Acquiring lexical knowledge for anaphora resolution. In Proc. of LREC 2002, pages 1220–1224. find our card
  16. J. Preiss. 2002. Anaphora resolution with word sense disambiguation. In Proc. of SENSEVAL-2, pages 143– 146. find
  17. W. M. Soon, H. T. Ng, and D. C. Y. Lim. 2001. A machine learning approach to coreference resolution of noun phrases. Computational Linguistics, 27(4):521– 544. find our card
  18. M. Strube, S. Rapp, and C. Müller. 2002. The influence of minimum edit distance on reference resolution. In Proc. of EMNLP 2002, pages 312–319. find
  19. M. Strube. 2002. NLP approaches to reference resolution. Tutorial notes, ACL’02. find

↑ back to top

This page in Markdown →