Automatic Processing of Large Corpora for the Resolution of Anaphora References
Summary
Hand-built selectional constraints are a standard way to resolve ambiguity, but writing them for broad domains is too costly. Dagan and Itai propose learning such constraints automatically from cooccurrence statistics in a large, unannotated corpus. In an acquisition phase, sentences from the Hansard corpus (Canadian parliamentary proceedings) are parsed with the PEG parser. Post-processing then normalizes the parses into subject-verb, verb-object and adjective-noun relations, and frequency counts are stored. In a disambiguation phase, each candidate antecedent of the pronoun "it" is substituted for the pronoun. A candidate is kept only if the resulting pattern occurs often enough in the corpus.
On 59 filtered test sentences, the method gave a decision in about two thirds of cases and was usually correct when it did. It often narrowed the choice to a single referent, which classic semantic constraints rarely achieve. The authors also found that parsing errors and ambiguity in the unannotated corpus did little harm. The work shows that statistics layered on conventional parsers can feasibly supply semantic knowledge for disambiguation.
Key points
- Statistics were gathered from roughly 28 million words of the Hansard corpus, which totals about 60M words. Only around 10,000 relevant sentences were actually parsed with the PEG parser.
- The test set had 59 occurrences of "it". Each had between 2 and 5 candidate referents, averaging 2.8, all taken from the same sentence. A pattern counted as approved at 5 or more occurrences.
- Coverage was 38/59 (64%), and accuracy on the covered cases was 33/38 (87%).
- In 18 cases, only the correct referent was approved. Traditional selectional constraints could have settled just 7 of these, consistent with Hobbs (1978).
- Semantically valid patterns turned up far more often in unambiguous constructions than in ambiguous ones. As a result, wrong patterns from bad parses seldom reached the threshold, and no lexical-sense confusion occurred.
Cited by
- Evaluation Tool for Rule-based Anaphora Resolution Methods
- Unsupervised Learning of Contextual Role Knowledge for Coreference Resolution
- Bootstrapping Path-Based Pronoun Resolution
- A Generic Anaphora Resolution Engine for Indian Languages
- Decentralized Entity-Level Modeling for Coreference Resolution
- Revisiting Selectional Preferences for Coreference Resolution
- Automatic Pronominal Anaphora Resolution in English Texts
- Pronominal and Sortal Anaphora Resolution for Biomedical Literature
- Multilingual Robust Anaphora Resolution
- Supervised Noun Phrase Coreference Research: The First Fifteen Years
- An Algorithm for Anaphora Resolution in Spanish Texts
- Functional Centering – Grounding Referential Coherence on Information Structure
- Mind the GAP: A Balanced Corpus of Gendered Ambiguous Pronouns
- A High-Performance Coreference Resolution System using a Constraint-based Multi-Agent Strategy
Cite
@inproceedings{dagan1990automatic,
title = {{Automatic Processing of Large Corpora for the Resolution of Anaphora References}},
author = {Dagan, Ido and Itai, Alon},
booktitle = {COLING 1990 Volume 3: Papers presented to the 13th International Conference on Computational Linguistics},
year = {1990},
doi = {10.3115/991146.991209},
url = {https://doi.org/10.3115/991146.991209}
}References
- Brown, P., Cocke, J., Della Pietra, S., Della Pietra, V., Jelinek, F., Mercer, R.L. and Roossin P.S., A statistical approach to language translation, COLING 1988. find our card
- Carbonell, J. G. and Brown, R. D. Anaphora resolution: A multi strategy approach, COLING I988. find
- Church, K. W., A stochastic parts program and noun phrase parser for unrestricted text, ACL Conf. on Applied NLP, 1988. find
- R. Grishman, L. Hirschman and Ngo Thanh Nhan, Discovery procedures for sublanguage selectional patterns initial experiments, Computational Linguislics, vol. 12,205-214, 1986. find
- ttobbs, J. R. Resolving pronoun references, Lingua, vol. 44,311-338, 1978. find
- K. Jensen, PEG 1986: A broad-coverage computational syntax of English, Technical Report, IBM T. J. Watson Research Center, 1986. find
- Jensen, K. PEGASOS: Deriving predicate-argument structures after a syntactic parse. Presented at the International Workshop on Parsing Technologies, Carnegie Mellon University, August 1989. find
- S. Lappin, I. Golan, M. Rimon, Computing grammatical fimctions from a configurational parse tree, Technical Report 88.268, IBM Israel Center of Science and Technology, 1989. find
This page in Markdown → https://md.research.elamur.ai/free/papers/dagan-1990-automatic-processing-large-corpora-resolution/