Elamur Research⌕
☰
home / papers / 2012 / Domain Adaptation for Coreference Resolution: An Adaptive Ensemble Approach

Domain Adaptation for Coreference Resolution: An Adaptive Ensemble Approach

Summary

Coreference resolution systems are usually trained and tested within a single domain, yet annotated documents for a particular domain of interest are often scarce while related domains may have plenty. The authors present what they describe as the first domain adaptation approach for this task: an adaptive ensemble. They build a pool of mention-pair base models for each source and target domain. These are SVM classifiers paired with closest-first or best-first clustering, run under varied regularization and threshold settings. A diagonal Mahalanobis-style distance between source and target documents is then learned. For each test document, the system selects the base models that performed best on its nearest labeled neighbours. This design lets the method optimize any chosen evaluation metric without a separate development set, tailor predictions to each document and automatically discard weak ensemble members. On six ACE 2005 domains with only five labeled target training documents, the method clearly beats within-domain baselines. A simplified version also outperforms other ensembles on MUC-6 and ACE 2005. The work shows that labeled data from neighbouring genres can substantially help low-resource coreference.

Key points

  • Base models: Mention pairs are represented by the 39 features of Rahman and Ng (2011b), extracted with Reconcile. These are enriched with RBF-like similarities to k-means centroids, which avoids the cost of a full rbf kernel. The classifiers are trained with Liblinear.
  • Document representation and inference: Each document is represented as the sum of its mention-pair feature vectors. Neighbourhood sizes k = 1, 3 and 5 were tested. In the cross-domain setting, the best k depended on the task.
  • Evaluation domains: The domain adaptation tasks cover six ACE 2005 domains: BC, BN, CTS, NW, UN and WL. Results are reported with B-CUBED and CEAF F-measures.
  • Main result: On average, the best adapted model beats the stronger within-domain baseline by 7.2% in B-CUBED F (71.2 vs 64.0) and by 3% in CEAF F (63.7 vs 60.7).
  • Value of cross-domain learning: Learning the cross-domain distance usually adds 1–3% over a grand ensemble that uses no such distance. Planned future work includes adaptation with no target labels at all.

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

7 papers of our library cite it (of the 140673 whose bibliographies we read); 5 of them have a page here, 2 not yet.

Cite

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

@inproceedings{yang2012domain,
  title = {{Domain Adaptation for Coreference Resolution: An Adaptive Ensemble Approach}},
  author = {Yang, Jian Bo and Mao, Qi and Xiang, Qiao Liang and Tsang, Ivor Wai-Hung and Chai, Kian Ming Adam and Chieu, Hai Leong},
  booktitle = {Proceedings of the 2012 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning},
  year = {2012},
  url = {https://aclanthology.org/D12-1068/}
}

References

29 references, 8 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Amit Bagga and Breck Baldwin. 1998. Entity-based cross-document coreferencing using the vector space model. In Proceedings of the 36th Annual Meeting of the Association for Computational Linguistics and 17th International Conference on Computational Linguistics - Volume 1, ACL’98, pages 79–85. find
  2. Christopher M. Bishop. 2007. Pattern Recognition and Machine Learning (Information Science and Statistics). Springer, 1st ed. 2006. corr. 2nd printing edition, October. find
  3. John Blitzer, Ryan McDonald, and Fernando Pereira. 2006. Domain adaptation with structural correspondence learning. In Proceedings of EMNLP, pages 120–128. find
  4. Leo Breiman. 1996. Bagging predictors. Machine Learning, 24(2):123–140, August. find our card
  5. Leo Breiman. 2001. Random forests. Machine Learning, 45(1):5–32, October. find
  6. Pascal Denis and Jason Baldridge. 2007. Joint determination of anaphoricity and coreference resolution using integer programming. In Proc HLT, pages 236– 243, Rochester, New York, April. find our card
  7. Rong-En Fan, Kai-Wei Chang, Cho-Jui Hsieh, Xiang-Rui Wang, and Chih-Jen Lin. 2008. LIBLINEAR: A library for large linear classification. Journal of Machine Learning Research, 9:1871–1874. find
  8. Thomas Finley and Thorsten Joachims. 2005. Supervised clustering with support vector machines. In Proc. ICML. find
  9. Yoav Freund and Robert E. Schapire. 1996. Experiments with a New Boosting Algorithm. In Proc. ICML, pages 148–156. find
  10. Honglei Guo, Huijia Zhu, Zhili Guo, Xiaoxun Zhang, Xian Wu, and Zhong Su. 2009. Domain adaptation with latent semantic association for named entity recognition. NAACL ’09, pages 281–289. find
  11. Jing Jiang and ChengXiang Zhai. 2007. Instance weighting for domain adaptation in NLP. In Proc. ACL, pages 264–271, Prague, Czech Republic, June. find
  12. Jing Jiang. 2009. Multi-task transfer learning for weakly-supervised relation extraction. In Proceedings of the Joint Conference of the 47th Annual Meeting of the ACL and the 4th International Joint Conference on Natural Language Processing of the AFNLP, pages 1012–1020, Suntec, Singapore, August. A… find
  13. Xiaoqiang Luo. 2005. On coreference resolution performance metrics. In Proceedings of the conference on Human Language Technology and Empirical Methods in Natural Language Processing, HLT ’05, pages 25– 32. find our card
  14. Art Munson, Claire Cardie, and Rich Caruana. 2005. Optimizing to arbitrary NLP metrics using ensemble selection. In Proc HLT and EMNLP, pages 539–546. find
  15. Vincent Ng and Claire Cardie. 2002. Improving machine learning approaches to coreference resolution. In Proc. ACL, pages 104–111. find our card
  16. Vincent Ng and Claire Cardie. 2003. Weakly supervised natural language learning without redundant views. In Proc. HLT-NAACL. find
  17. Vincent Ng. 2005. Machine learning for coreference resolution: From local classification to global ranking. In Proceedings of the ACL, pages 157–164. find our card
  18. Sinno Jialin Pan and Qiang Yang. 2010. A survey on transfer learning. IEEE Transactions on Knowledge and Data Engineering, 22(10):1345–1359, October. find
  19. Wenbo Pang and Xiaozhong Fan. 2009. Chinese coreference resolution with ensemble learning. In Proc. PACIIA, pages 236–243. find
  20. Altaf Rahman and Vincent Ng. 2011a. Ensemble-based coreference resolution. In Proceedings of IJ-CAI, pages 1884–1889. find
  21. Altaf Rahman and Vincent Ng. 2011b. Narrowing the modeling gap: A cluster-ranking approach to coreference resolution. JAIR, 1:469–52. find
  22. William M. Rand. 1971. Objective criteria for the evaluation of clustering methods. Journal of the American Statistical Association, 66(336):pp. 846–850. find
  23. W. M. Soon, H. T. Ng, and D. C. Y. Lim. 2001. A machine learning approach to coreference resolution of noun phrases. Computational Linguistics,, pages 521– 544. find our card
  24. Veselin Stoyanov, Claire Cardie, Nathan Gilbert, Ellen Riloff, David Buttler, and David Hysom. 2010. Coreference resolution with reconcile. In Proc. ACL, pages 156–161. find our card
  25. Songbo Tan, Yuefen Wang, Gaowei Wu, and Xueqi Cheng. 2008. Using unlabeled data to handle domain-transfer problem of semantic detection. In Proceedings of the 2008 ACM symposium on Applied computing, SAC ’08, pages 896–903. find
  26. S. Vemulapalli, X. Luo, J.F.Pitrelli, and I. Zitouni. 2009. classifier combination applied to coreference resolution. In NAACL HLT Student Rsearch Workshop. find
  27. Marc Vilain, John Burger, John Aberdeen, Dennis Connolly, and Lynette Hirschman. 1995. A model-theoretic coreference scoring scheme. In Proceedings of the 6th conference on Message understanding, MUC6 ’95, pages 45–52. find our card
  28. Liu Yang and Rong Jin. 2006. Distance Metric Learning: A Comprehensive Survey. Technical report, Department of Computer Science and Engineering, Michigan State University. find
  29. Chun-Nam John Yu and Thorsten Joachims. 2009. Learning structural SVMs with latent variables. In Proc. ICML, pages 1169–1176, New York, NY, USA. find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/papers/yang-2012-domain-adaptation-coreference-resolution-adaptive/