Elamur Research⌕
☰
home / papers / 2007 / Joint Determination of Anaphoricity and Coreference Resolution using Integer Programming

Joint Determination of Anaphoricity and Coreference Resolution using Integer Programming

Summary

Pairwise coreference systems only decide anaphoricity implicitly. As a result, they may link non-anaphoric mentions to antecedents or leave genuine anaphors unresolved. Earlier fixes placed a separate anaphoricity classifier in front of the coreference model as a filter, but such cascades depend on a carefully tuned threshold. Denis and Baldridge instead treat coreference as an integer linear programming (ILP) problem. Two maximum entropy classifiers, one scoring mention pairs and one scoring whether a mention is anaphoric, supply probabilities that serve as weights in the ILP objective. Linear constraints then force the two sets of decisions to be consistent. The authors first build a coreference-only ILP (COREF-ILP) and then the joint model (JOINT-ILP). They test on the three genres of ACE Phase 2 (BNEWS, NPAPER, NWIRE), using true ACE mentions and the Vilain et al. (1995) metric. The joint model improves f-score by 3.7–5.3% over the pairwise baseline, and all differences are significant. The result matters because global inference can combine independently trained local models quickly and without threshold tuning.

Key points

  • Base models are maximum entropy classifiers trained with TADM (Gaussian prior, variance 1000). Training pairs follow Soon et al. (2001), and preprocessing uses the OpenNLP Toolkit.
  • Baseline f-scores for COREF-PAIRWISE are 63.9 on BNEWS, 67.6 on NPAPER and 63.8 on NWIRE. The filter cascade (AC-CASCADE) scores lower on BNEWS and NPAPER and only marginally higher on NWIRE.
  • The anaphoricity classifier reaches about 80.2% average accuracy. Even with perfect linkage (ORACLE-LINK), its errors cap a cascade at roughly 80–83 f-score.
  • JOINT-ILP yields f-score gains of 5.3%, 4.9% and 3.7% across the three ACE datasets, with significance tested by paired t-tests (p < .05).
  • Planned future work includes transitivity constraints, partition-level constraints and linguistic constraints such as pronoun restrictions.

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

73 papers of our library cite it (of the 140673 whose bibliographies we read); 41 of them have a page here, 32 not yet.

Cite

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

@inproceedings{denis2007joint,
  title = {{Joint Determination of Anaphoricity and Coreference Resolution using Integer Programming}},
  author = {Denis, Pascal and Baldridge, Jason},
  booktitle = {Human Language Technologies 2007: The Conference of the North American Chapter of the Association for Computational Linguistics; Proceedings of the Main Conference},
  year = {2007},
  url = {https://aclanthology.org/N07-1030/}
}

References

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

  1. Regina Barzilay and Mirella Lapata. 2006. Aggregation via set partitioning for natural language generation. In Proceedings of the HLT/NAACL, pages 359–366, New York, NY. find
  2. A. Berger, S. Della Pietra, and V. Della Pietra. 1996. A maximum entropy approach to natural language processing. Computational Linguistics, 22(1):39–71. find our card
  3. A. Kehler, D. Appelt, L. Taylor, and A. Simma. 2004. The (non)utility of predicate-argument frequencies for pronoun interpretation. In Proceedings of HLT/NAACL, pages 289–296. find
  4. Andrew Kehler. 1997. Probabilistic coreference in information extraction. In Proceedings of EMNLP, pages 163–173. find our card
  5. Xiaoqiang Luo, Abe Ittycheriah, Hongyan Jing, Nanda Kambhatla, , and Salim Roukos. 2004. A mention-synchronous coreference resolution algorithm based on the Bell tree. In Proceedings of the ACL. find our card
  6. Robert Malouf. 2002. A comparison of algorithms for maximum entropy parameter estimation. In Proceedings of the Sixth Workshop on Natural Language Learning, pages 49–55, Taipei, Taiwan. find our card
  7. Andrew McCallum and Ben Wellner. 2004. Conditional models of identity uncertainty with application to noun coreference. In Proceedings of NIPS. find
  8. Joseph F. McCarthy and Wendy G. Lehnert. 1995. Using decision trees for coreference resolution. In Proceedings of IJCAI, pages 1050–1055. find
  9. Thomas Morton. 1999. Using coreference for question answering. In Proceedings of ACL Workshop on Coreference and Its Applications. find our card
  10. Thomas Morton. 2000. Coreference for NLP applications. In Proceedings of ACL, Hong Kong. find
  11. Vincent Ng and Claire Cardie. 2002a. Identifying anaphoric and non-anaphoric noun phrases to improve coreference resolution. In Proceedings of COLING. find
  12. Vincent Ng and Claire Cardie. 2002b. Improving machine learning approaches to coreference resolution. In Proceedings of ACL, pages 104–111. find our card
  13. Vincent Ng. 2004. Learning noun phrase anaphoricity to improve coreference resolution: Issues in representation and optimization. In Proceedings of ACL. find
  14. Vincent Ng. 2005. Machine learning for coreference resolution: From local classification to global ranking. In Proceedings of ACL. find our card
  15. Dan Roth and Wen-tau Yih. 2004. A linear programming formulation for global inference in natural language tasks. In Proceedings of CoNLL. find our card
  16. Dan Roth and Wen-tau Yih. 2005. Integer linear programming inference for conditional random fields. In Proceedings of ICML, pages 737–744. find
  17. W. Soon, H. Ng, and D. Lim. 2001. A machine learning approach to coreference resolution of noun phrases. Computational Linguistics, 27(4):521–544. find our card
  18. Marc Vilain, John Burger, John Aberdeen, Dennis Connolly, and Lynette Hirschman. 1995. A model-theoretic coreference scoring scheme. In Proceedings fo the 6th Message Understanding Conference (MUC-6), pages 45–52, San Mateo, CA. Morgan Kaufmann. find our card

↑ back to top

This page in Markdown →