Bootstrapping Path-Based Pronoun Resolution

Proceedings of the 21st International Conference on Computational Linguistics and 44th Annual Meeting of the ACL, pages 33–40, Sydney, July 2006. c⃝2006 Association for Computational Linguistics Bootstrapping Path-Based Pronoun Resolution Shane Bergsma Department of Computing Science University of Alberta Edmonton, Alberta, Canada, T6G 2E8 bergsma@cs.ualberta.ca Dekang Lin Google, Inc. 1600 Amphitheatre Parkway, Mountain View, California, 94301 lindek@google.com Abstract We present an approach to pronoun reso- lution based on syntactic paths. Through a simple bootstrapping procedure, we learn the likelihood of coreference between a pronoun and a candidate noun based on the path in the parse tree between the two en- tities. This path information enables us to handle previously challenging resolution instances, and also robustly addresses tra- ditional syntactic coreference constraints. Highly coreferent paths also allow mining of precise probabilistic gender/number in- formation. We combine statistical knowl- edge with well known features in a Sup- port Vector Machine pronoun resolution classifier. Significant gains in performance are observed on several datasets. 1 Introduction Pronoun resolution is a difficult but vital part of the overall coreference resolution task. In each of the following sentences, a pronoun resolution system must determine what the pronoun his refers to: (1) John needs his friend. (2) John needs his support. In (1), John and his corefer. In (2), his refers to some other, perhaps previously evoked entity. Traditional pronoun resolution systems are not de- signed to distinguish between these cases. They lack the specific world knowledge required in the second instance – the knowledge that a person does not usually explicitly need his own support. We collect statistical path-coreference informa- tion from a large, automatically-parsed corpus to address this limitation. A dependency path is de- fined as the sequence of dependency links between two potentially coreferent entities in a parse tree. A path does not include the terminal entities; for example, “John needs his support” and “He needs their support” have the same syntactic path. Our algorithm determines that the dependency path linking the Noun and pronoun is very likely to con- nect coreferent entities for the path “Noun needs pronoun’s friend,” while it is rarely coreferent for the path “Noun needs pronoun’s support.” This likelihood can be learned by simply count- ing how often we see a given path in text with an initial Noun and a final pronoun that are from the same/different gender/number classes. Cases such as “John needs her support” or “They need his support” are much more frequent in text than cases where the subject noun and pronoun termi- nals agree in gender/number. When there is agree- ment, the terminal nouns are likely to be corefer- ent. When they disagree, they refer to different en- tities. After a sufficient number of occurrences of agreement or disagreement, there is a strong sta- tistical indication of whether the path is coreferent (terminal nouns tend to refer to the same entity) or non-coreferent (nouns refer to different entities). We show that including path coreference in- formation enables significant performance gains on three third-person pronoun …

Content truncated to 500 words