Analysis of Syntax-Based Pronoun Resolution Methods

Summary

The paper addresses two gaps in pronoun resolution research. First, the Centering-based BFP algorithm (Brennan et al., 1987) cannot resolve pronouns incrementally and can become very expensive to compute. Second, few large-scale empirical comparisons of resolution methods existed.

Tetreault proposes Left-Right Centering (LRC), which follows Centering Theory constraints. For each pronoun, LRC first looks for an antecedent among the entities already processed in the current sentence. Only if none is found does it scan earlier Cf-lists from left to right. It never builds or filters the anchor sets that BFP relies on.

LRC, Hobbs' naive algorithm, a syntax-only version of Strube's S-list and BFP were all implemented and run on a parsed, annotated Penn Treebank subset from Charniak and Ge (1998). Hobbs and LRC performed best, with a statistically insignificant gap between them. BFP came last.

The study matters because it was the first automated, common-corpus comparison of these methods. It also offers LRC as a cognitively more plausible substitute for BFP that does not lose accuracy.

Key points

Data: about 3900 utterances from 195 newspaper articles, containing 2026 pronouns. Removing quoted text left 1696 pronouns for the main evaluation.
Methods that search all previous Cf-lists:
- Hobbs: 72.8% correct (1234 pronouns)
- LRC-N: 72.4% correct (1228 pronouns)
- Strube-N: 68.8% correct
Methods restricted to the previous utterance:
- LRC-1: 71.2%
- Strube-1: 66.0%
- BFP: 56.7%
- Most-recent-NP baseline: 28.6%
Why Hobbs and LRC win: over 71% of pronouns have their antecedent in the same sentence, which favours methods that check the current sentence first. BFP was hurt by its preference for antecedents in earlier sentences.
Including quoted text: across all 2026 pronouns, LRC scored 70.4% and Hobbs 69.8%. BFP's cost problem is illustrated by one example sentence that produces 36864 anchors.

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

9 papers of our library cite it (of the 140673 whose bibliographies we read); 3 of them have a page here, 6 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.3115/1034678.1034688

@inproceedings{tetreault1999analysis,
  title = {{Analysis of Syntax-Based Pronoun Resolution Methods}},
  author = {Tetreault, Joel R.},
  booktitle = {Proceedings of the 37th Annual Meeting of the Association for Computational Linguistics},
  year = {1999},
  doi = {10.3115/1034678.1034688},
  url = {https://doi.org/10.3115/1034678.1034688}
}

References

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

Susan E. Brennan, Marilyn W. Friedman, and Carl J. Pollard. 1987. A centering approach to pronouns. In Proceedings, 25th Annual Meeting of the ACL, pages 155-162. find
Niyu Ge, John Hale, and Eugene Charniak. 1998. A statistical approach to anaphora resolution. Proceedings of the Sixth Workshop on Very Large Corpora. find our card
Barbara J. Grosz, Aravind K. Joshi, and Scott Weinstein. 1995. Centering: A framework for modeling the local coherence of discourse. Computational Linguistics, 21 (2):203-226. find our card
Jerry R. Hobbs. 1977. Resolving pronoun references. Lingua, 44:311-338. find our card
Megumi Kameyama. 1986. Intrasentential centering: A case study. In Centering Theory in Discourse. find
Andrew Kehler. 1997. Current theories of centering for pronoun interpretation: A critical evaluation. Computational Linguistics, 23(3):467-475. find
Mitchell P. Marcus, Beatrice Santorini, and Mary Ann Marcinkiewicz. 1993. Building a large annotated corpus of english: The penn treebank. Computational Lingusitics, 19(2):313-330. find our card
Michael Strube and Udo Hahn. 1996. Functional centering. In Association for Computational Lingusitics, pages 270-277. find
Michael Strube. 1998. Never look back: An alternative to centering. In Association for Computational Lingusitics, pages 1251-1257. find
Marilyn A. Walker. 1989. Evaluating discourse processing algorithms. In Proceedings, 27th Annual Meeting of the Association for Compuational Linguisites, pages 251-261. find

This page in Markdown → https://md.research.elamur.ai/free/en/papers/tetreault-1999-analysis-syntax-based-pronoun-resolution/

📄 Analysis of Syntax-Based Pronoun Resolution Methods