Elamur Research⌕
☰
home / papers / 2013 / A Constrained Latent Variable Model for Coreference Resolution

A Constrained Latent Variable Model for Coreference Resolution

Summary

Coreference resolution clusters noun-phrase mentions according to which real-world entity each one denotes. Earlier systems either relied on pairwise mention scores trained separately from the clustering step, or used richer mention–cluster models that need heuristics to stay tractable. The authors introduce the Latent Left-Linking Model (L3M). In this structured prediction model, each mention attaches to its highest-scoring preceding antecedent, or to a dummy mention if none qualifies, and the clustering is the transitive closure of these links. Training and decoding are coupled through a latent structural SVM objective, which is optimized with fast stochastic gradient descent and uses a decomposable pairwise loss related to the Rand index. A probabilistic extension, PL3M, considers several left-links at once, controlled by a temperature-like parameter. The constrained variants CL3M and CPL3M add domain-knowledge constraints. On ACE 2004 and OntoNotes-5.0, the models outperform correlational clustering, latent spanning forests, IllinoisCoref and the Stanford rule-based system. This shows that a well-trained mention-pair model can match more complex entity-level approaches.

Key points

  • Datasets: ACE 2004 and OntoNotes-5.0. Results are reported with MUC, B3 (BCUB) and CEAF-e metrics and their average.
  • CL3M uses five constraints: SameSpan, SameDetNom, SameProperName, ModifierMismatch and PropertyMismatch.
  • On OntoNotes-5.0 with predicted mentions, the constraints raise average F1 from 62.30 (L3M) to 63.59 (CL3M), a gain of 1.29. Of that gain, 1.19 comes from the four must-link constraints, which help because L3M has higher precision (78.38% B3) than recall (67.96%).
  • Considering multiple links (γ = 0.2 vs. γ = 0) adds about 0.7 points on ACE 2004. With noisy predicted mentions, γ = 0 worked best.
  • In the reported comparison, correlational clustering with exact ILP inference beat the spanning forest approach. Constraints helped less on ACE because its mentions are shorter phrases.

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

27 papers of our library cite it (of the 140673 whose bibliographies we read); 13 of them have a page here, 14 not yet.

Cite

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

@inproceedings{chang2013constrained,
  title = {{A Constrained Latent Variable Model for Coreference Resolution}},
  author = {Chang, Kai-Wei and Samdani, Rajhans and Roth, Dan},
  booktitle = {Proceedings of the 2013 Conference on Empirical Methods in Natural Language Processing},
  year = {2013},
  doi = {10.18653/v1/d13-1057},
  url = {https://doi.org/10.18653/v1/d13-1057}
}

References

49 references, 25 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. A. Bagga and B. Baldwin. 1998. Algorithms for scoring coreference chains. In In The First International Conference on Language Resources and Evaluation Workshop on Linguistics Coreference. find our card
  2. M. Bansal and D. Klein. 2012. Coreference semantics from web features. In Proceedings of ACL, Jeju Island, South Korea, July. find our card
  3. N. Bansal, A. Blum, and S. Chawla. 2002. Correlation clustering. In Proceedings of the 43rd Symposium on Foundations of Computer Science. find
  4. E. Bengtson and D. Roth. 2008. Understanding the value of features for coreference resolution. In EMNLP, 10. find our card
  5. K.-W. Chang, R. Samdani, A. Rozovskaya, N. Rizzolo, M. Sammons, and D. Roth. 2011. Inference protocols for coreference resolution. In CoNLL Shared Task. find our card
  6. K.-W. Chang, R. Samdani, A. Rozovskaya, M. Sammons, and D. Roth. 2012a. Illinois-coref: The UI system in the CoNLL-2012 Shared Task. In CoNLL Shared Task. find
  7. M. Chang, L. Ratinov, and D. Roth. 2012b. Structured learning with constrained conditional models. Machine Learning, 88(3):399–431, 6. find
  8. C. Chen and V. Ng. 2012. Combining the best of two worlds: A hybrid approach to multilingual coreference resolution. In Joint Conference on EMNLP and CoNLL - Shared Task. find our card
  9. J. Clarke and M. Lapata. 2006. Constraint-based sentence compression: An integer programming approach. In Proceedings of the Annual Meeting of the Association for Computational Linguistics (ACL), pages 144–151, Sydney, Australia, July. ACL. find
  10. A. Culotta, M. Wick, R. Hall, and A. McCallum. 2007. First-order probabilistic models for coreference resolution. In HLT/NAACL. find our card
  11. P. Denis and J. Baldridge. 2007. Joint determination of anaphoricity and coreference resolution using integer programming. In Proceedings of the Annual Meeting of the North American Association of Computational Linguistics (NAACL). find our card
  12. P. Denis and J. Baldridge. 2008. Specialized models and ranking for coreference resolution. In EMNLP, pages 660–669. find our card
  13. G. Durrett, D. Hall, and D. Klein. 2013. Decentralized entity-level modeling for coreference resolution. In Proceedings of ACL, August. find our card
  14. E. R. Fernandes, C. N. dos Santos, and R. L. Milidiú. 2012. Latent structure perceptron with feature induction for unrestricted coreference resolution. In Joint Conference on EMNLP and CoNLL - Shared Task. find our card
  15. T. Finley and T. Joachims. 2005. Supervised clustering with support vector machines. In Proceedings of the International Conference on Machine Learning (ICML). find
  16. K. Ganchev, J. Graça, J. Gillenwater, and B. Taskar. 2010. Posterior regularization for structured latent variable models. Journal of Machine Learning Research. find
  17. K. Gimpel and N. A. Smith. 2010. Softmax-margin CRFs: Training log-linear models with cost functions. In NAACL. find
  18. A. Haghighi and D. Klein. 2010. Coreference resolution in a modular, entity-centered model. In NAACL. find our card
  19. H. Ji and R. Grishman. 2011. Knowledge base population: successful approaches and challenges. In Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies - Volume 1. find
  20. T. Koo, A. M. Rush, M. Collins, T. Jaakkola, and D. Sontag. 2010. Dual decomposition for parsing with non-projective head automata. In EMNLP. find
  21. H. Lee, Y. Peirsman, A. Chang, N. Chambers, M. Surdeanu, and D. Jurafsky. 2011. Stanford’s multi-pass sieve coreference resolution system at the conll-2011 shared task. In Proceedings of the CoNLL-2011 Shared Task. find our card
  22. H. Lee, A. Chang, Y. Peirsman, N. Chambers, M. Surdeanu, and D. Jurafsky. 2013. Deterministic coreference resolution based on entity-centric, precision-ranked rules. Computational Linguistics, 39(4). find our card
  23. X. Luo. 2005. On coreference resolution performance metrics. In EMNLP. find our card
  24. S. Martschat, J. Cai, S. Broscheit, É. Mújdricza-Maydt, and M. Strube. 2012. A multigraph model for coreference resolution. In Joint Conference on EMNLP and CoNLL - Shared Task, July. find
  25. A. McCallum and B. Wellner. 2003. Toward conditional models of identity uncertainty with application to proper noun coreference. In The Conference on Advances in Neural Information Processing Systems (NIPS). find
  26. R. Mihalcea and A. Csomai. 2007. Wikify!: linking documents to encyclopedic knowledge. In CIKM. find
  27. V. Ng and C. Cardie. 2002. Improving machine learning approaches to coreference resolution. In ACL. find our card
  28. Vincent Ng. 2005. Supervised ranking for pronoun resolution: Some recent improvements. In AAAI, pages 1081–1086. find
  29. NIST. 2004. The ACE evaluation plan. find
  30. D. Pascal and J. Baldridge. 2009. Global joint models for coreference resolution and named entity classification. In Procesamiento del Lenguaje Natural. find
  31. P. Pletscher, C. S. Ong, and J. M. Buhmann. 2010. Entropy and margin maximization for structured output learning. In ECML PKDD. find
  32. S. Pradhan, L. Ramshaw, M. Marcus, M. Palmer, R. Weischedel, and N. Xue. 2011. Conll-2011 shared task: Modeling unrestricted coreference in ontonotes. In CoNLL. find our card
  33. S. Pradhan, A. Moschitti, N. Xue, O. Uryupina, and Y. Zhang. 2012. CoNLL-2012 shared task: Modeling multilingual unrestricted coreference in OntoNotes. In CoNLL 2012. find our card
  34. M. Sammons Y. Tu V. Vydiswaran Q. Do, D. Roth. 2009. Robust, light-weight approaches to compute lexical similarity. Technical report. find
  35. K. Raghunathan, H. Lee, S. Rangarajan, N. Chambers, M. Surdeanu, D. Jurafsky, and C. Manning. 2010. A multi-pass sieve for coreference resolution. In EMNLP. find our card
  36. A. Rahman and V. Ng. 2011a. Coreference resolution with world knowledge. In ACL, pages 814–824. find our card
  37. A. Rahman and V. Ng. 2011b. Ensemble-based coreference resolution. In IJCAI. find
  38. A. Rahman and V. Ng. 2011c. Narrowing the modeling gap: a cluster-ranking approach to coreference resolution. JAIR. find
  39. W.M. Rand. 1971. Objective criteria for the evaluation of clustering methods. Journal of the American Statistical Association, 66(336):846–850. find
  40. L. Ratinov and D. Roth. 2012. Learning-based multi-sieve co-reference resolution with knowledge. In EMNLP. find our card
  41. L. Ratinov, D. Roth, D. Downey, and M. Anderson. 2011. Local and global algorithms for disambiguation to wikipedia. In Proc. of the Annual Meeting of the Association for Computational Linguistics (ACL). find
  42. Dan Roth and Wen Tau Yih. 2004. A linear programming formulation for global inference in natural language tasks. In Proceedings of CoNLL-04, pages 1–8. find our card
  43. R. Samdani, M. Chang, and D. Roth. 2012. Unified expectation maximization. In NAACL. find
  44. W. M. Soon, H. T. Ng, and D. C. Y. Lim. 2001. A machine learning approach to coreference resolution of noun phrases. Comput. Linguist. find our card
  45. V. Stoyanov and J. Eisner. 2012. Easy-first coreference resolution. In COLING, pages 2519–2534. find our card
  46. V. Stoyanov, N. Gilbert, C. Cardie, and E. Riloff. 2009. Conundrums in noun phrase coreference resolution: making sense of the state-of-the-art. In ACL. find our card
  47. M. Vilain, J. Burger, J. Aberdeen, D. Connolly, and L. Hirschman. 1995. A model-theoretic coreference scoring scheme. In Proceedings of the 6th conference on Message understanding. find our card
  48. C. Yu and T. Joachims. 2009. Learning structural svms with latent variables. In Proceedings of the International Conference on Machine Learning (ICML). find
  49. A. L. Yuille and A. Rangarajan. 2003. The concave-convex procedure. Neural Computation, 15(4). find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/chang-2013-constrained-latent-variable-model-coreference/