Corpus-Based Identification of Non-Anaphoric Noun Phrases

Summary

Coreference resolvers usually look for an antecedent for every definite noun phrase. Many such NPs, which the authors call "existential", need none: world knowledge or their own modifiers already make the referent clear, as with "the White House". Bean and Riloff build an automatic, corpus-driven filter that flags these non-anaphoric definite NPs before resolution. It needs no hand-coded lists and no full parsing. Syntactic heuristics are combined with three resources learned from 1,600 MUC-4 terrorism articles:

definite NPs from first sentences (S1)
generalized Existential Head Patterns (EHP)
a Definite-Only (DO) list of NPs that practically never appear with indefinite determiners

A "vaccine" step uses definiteness probabilities to discard unreliable S1/EHP entries. On 50 hand-annotated test texts, the tuned system reached 78% recall and 87% precision, slightly beating the earlier Vieira & Poesio system. Precision matters most for a pre-filter, and about two thirds of test definite NPs were independent existentials. The method could therefore make coreference resolution faster and more accurate while staying portable across domains.

Key points

The test set had 1,001 definite NPs:
- 48% syntactically independent existential
- 15% semantically independent existential
- 9% associative existential
- 28% referential
The baseline (label everything existential) gives 100% recall and 72% precision. Syntactic heuristics alone (seven rule-in, two rule-out) reach 43% recall at roughly 92–93% precision.
Training produced 297 EHPs and a DO list of 65 head nouns plus 321 full NPs. The DO thresholds were at least 5 occurrences, with 100% definite use for head nouns and 75% for full NPs.
All methods combined (heuristics + S1 + EHP + DO) gave 81.7% recall and 82.2% precision. Vaccine Va (upper threshold 70%, lower 25%) moved this to 77.7% recall and 86.6% precision. Vb (upper 50%) gave 79.1% recall and 84.5% precision.
Vieira & Poesio reported 74% recall and 85% precision. Scored only on independent existentials, this system reaches 81.8% recall and 85.6% precision with Va.

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

23 papers of our library cite it (of the 140673 whose bibliographies we read); 12 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. DOI 10.3115/1034678.1034737

@inproceedings{bean1999corpus,
  title = {{Corpus-Based Identification of Non-Anaphoric Noun Phrases}},
  author = {Bean, David L. and Riloff, Ellen},
  booktitle = {Proceedings of the 37th Annual Meeting of the Association for Computational Linguistics},
  year = {1999},
  doi = {10.3115/1034678.1034737},
  url = {https://doi.org/10.3115/1034678.1034737}
}

References

10 references, 1 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

James Allen. 1995. Natural Language Understanding. Benjamin/Cummings Press, Redwood City, CA. find
Chinatsu Aone and Scott William Bennett. 1996. Applying Machine Learning to Anaphora Resolution. In Connectionist, Statistical, and Symbolic Approaches to Learning for Natural Language Understanding, pages 302-314. Springer-Verlag, Berlin. find
Andrew Kehler. 1997. Probabilistic coreference in information extraction. In Proceedings of the Second Conference on Empirical Methods in Natural Language Processing (EMNLP-97). find our card
Christopher Kennedy and Branimir Boguraev. 1996. Anaphor for everyone: Pronomial anaphora resolution without a parser. In Proceedings of the 16th International Conference on Computational Linguistics (COLING-96). ~Case sensitive text can have a significant positive effect on performance because it h… find
Shalom Lappin and Herbert J. Leass. 1994. An algorithm for pronomial anaphora resolution. Computational Linguistics, 20(4):535-561. find
Joseph F. McCarthy and Wendy G. Lehnert. 1995. Using Decision Trees for Coreference Resolution. find
In Proceedings of the l~th International Joint Conference on Artificial Intelligence (IJCAI-95), pages 1050-1055. find
Ellen F. Prince. 1981. Toward a taxonomy of given-new information. In Peter Cole, editor, Radical Pragmatics, pages 223-255. Academic Press. find
Brian Roark and Eugene Charniak. 1998. Noun-phrase co-occurence statistics for semi-automatic semantic lexcon construction. In Proceedings of the 36th Annual Meeting of the Association for Computational Linguistics. find
R. Vieira and M. Poesio. 1997. Processing definite descriptions in corpora. In S. Botley and M. McEnery, editors, Corpus-based and Computational Approaches to Discourse Anaphora. UCL Press. find

↑ back to top

📄 Corpus-Based Identification of Non-Anaphoric Noun Phrases