Elamur Research
☰
home / papers / 2013 / Global Inference for Bridging Anaphora Resolution

Global Inference for Bridging Anaphora Resolution

What it is about

The paper addresses antecedent selection for bridging anaphora with no limits on anaphor or relation types. The usual pairwise model classifies each anaphor and candidate in isolation, with a fixed window. The authors use a Markov logic network instead. Local formulas use a rich feature set (preposition patterns, semantic relatedness, salience). Global formulas add constraints that anaphors in a text often share one antecedent, and that a salient entity is preferred. The corpus is 663 bridging NPs in OntoNotes (ISNotes), evaluated by 10-fold cross-validation on documents. The global model reaches 41.32% accuracy, against 36.35% for the best pairwise model. The gain is largest for antecedents three or more sentences away: 35.44% against 16.46%. The global model also reaches 91% of NP antecedents as candidates, against 77% for the local model.

Ideas

For Mova

The chapter is languages/en/35-reference-and-coreference.md. The code is coref/src/sieve.rs and coref/src/mention.rs, with the candidate salience in the chain code. The measure is accuracy by distance bucket, run twice for determinism. The Markov logic network is not for Mova. The global constraints can be written as rules. ISNotes uses OntoNotes data, so treat it as evaluation only.

Library summary

The Research library's own summary of the paper, in our words — not the paper's text.

Written for the Research library (claude-opus-5-5) from the paper text, in our own words: the license "CC BY-NC-SA 3.0 (ACL Anthology, before 2016)" does not allow redistributing the text itself. Original — aclanthology.org/N13-1111.pdf

Summary

Bridging anaphora point to an earlier entity that is related but not identical, as when "the windows" refers back to a building. Resolving them helps capture entity coherence and may support textual entailment. Earlier antecedent-selection systems handled only part-of relations or definite NP anaphors. Hou, Markert and Strube address antecedent selection for bridging without these restrictions. They use their corpus of 50 WSJ texts from OntoNotes, which contains 663 annotated bridging anaphors. They first build a local pairwise mention-entity model with new semantic, syntactic and salience features. They then recast the task as a global model in Markov logic networks, implemented with thebeast. This model encodes the tendency of related anaphors to share an antecedent and favours globally salient entities. It also draws candidates from the whole document instead of a fixed sentence window. The best variant, MLN model II, reaches 41.32% accuracy. This beats the strongest pairwise model (36.35%) and baselines reimplementing prior work (under 19%). The authors present it as the first resolver for unrestricted bridging in a realistic setting.

Key points

  • Corpus: about 11,000 NPs in total, of which 663 are bridging anaphors.
    • Only 38% of the anaphors are definite NPs.
    • 6% have non-NP antecedents, mostly verb phrases.
    • Just 14% involve part-of/attribute-of relations and 7% set relations.
    • Annotators agreed on antecedent choice about 80% of the time.
  • Why a fixed window fails: 71% of NP antecedents lie within 2 sentences of the anaphor and 81% within 5. Even a 2-sentence window yields a 1:17 ratio of positive to negative training instances. Searching the whole document makes 91% of NP antecedents reachable, compared with 77% for pairwise model III.
  • Accuracy (Table 4):
    • Baselines: 18.85 and 18.40
    • Pairwise models I/II/III: 29.11, 33.94 and 36.35
    • MLN models I/II: 35.60 and 41.32
  • Where the global model helps: for antecedents 3 or more sentences away, MLN II resolves 35.44% versus 16.46% for pairwise III. It reaches 54% on "sibling" anaphors that share an antecedent, but only 24% on non-siblings.
  • Remaining errors:
    • Non-NP antecedents are never resolved, since only NPs are extracted as candidates.
    • Unusual, context-specific relations are missed.
    • The model struggles when several semantically plausible antecedents compete, partly because it uses only head-noun knowledge.

Topics: Coreference resolution

Cite

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

@inproceedings{hou2013global,
  title = {{Global Inference for Bridging Anaphora Resolution}},
  author = {Hou, Yufang and Markert, Katja and Strube, Michael},
  booktitle = {Proceedings of the 2013 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies},
  year = {2013},
  url = {https://aclanthology.org/N13-1111/}
}

References

34 references, 3 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. 2008. Modeling local coherence: An entity-based approach. Computational Linguistics, 34(1):1–34. find our card
  2. Thorsten Brants and Alex Franz. 2006. Web 1t 5-gram version 1. LDC2006T13, Philadelphia, Penn.: Linguistic Data Consortium. find
  3. Aoife Cahill and Arndt Riester. 2012. Automatically acquiring fine-grained information status distinctions in German. In Proceedings of the SIGdial 2012 Conference: The 13th Annual Meeting of the Special Interest Group on Discourse and Dialogue, Seoul, Korea, 5–6 July 2012, pages 232–236. find
  4. Tommaso Caselli and Irina Prodanof. 2006. Annotating bridging anaphors in Italian: In search of reliability. In Proceedings of the 5th International Conference on Language Resources and Evaluation, Genoa, Italy, 22–28 May 2006. find
  5. Herbert H. Clark. 1975. Bridging. In Proceedings of the Conference on Theoretical Issues in Natural Language Processing, Cambridge, Mass., June 1975, pages 169– 174. find
  6. Pedro Domingos and Daniel Lowd. 2009. Markov Logic: An Interface Layer for Artificial Intelligence. Morgan Claypool Publishers. find
  7. Angela Fahrni and Michael Strube. 2012. Jointly disambiguating and clustering concepts and entities with Markov logic. In Proceedings of the 24th International Conference on Computational Linguistics, Mumbai, India, 8–15 December 2012, pages 815–832. find
  8. Kari Fraurud. 1990. Definiteness and the processing of noun phrases in natural discourse. Journal of Semantics, 7:395–433. find
  9. Claire Gardent and Hélène Manuélian. 2005. Création d’un corpus annoté pour le traitement des descriptions définies. Traitement Automatique des Langues, 46(1):115–140. find
  10. Matthew Gerber and Joyce Chai. 2012. Semantic role labeling of implicit arguments for nominal predicates. Computational Linguistics, 38(4):756–798. find
  11. Jeanette K. Gundel, Nancy Hedberg, and Ron Zacharski. 1993. Cognitive status and the form of referring expressions in discourse. Language, 69:274–307. find
  12. Donna Harman and Mark Liberman. 1993. TIPSTER Complete. LDC93T3A, Philadelphia, Penn.: Linguistic Data Consortium. Iorn Korzen and Matthias Buch-Kromann. Anaphoric relations in the Copenhagen dependency treebanks. In S. Dipper and H. Zinsmeister, editors, Corpus-based Investigations of Pragmatic and Discourse Phenomena, volume 3 of Bochumer Linguistische Arbeitsberichte, pages 83–98. University of Bochum, Bochum, Germany. find
  13. Emmanuel Lassalle and Pascal Denis. 2011. Leveraging different meronym discovery methods for bridging resolution in French. In Proceedings of the 8th Discourse Anaphora and Anaphor Resolution Colloquium (DAARC 2011), Faro, Algarve, Portugal, 6–7 October 2011, pages 35–46. find
  14. Katja Markert, Malvina Nissim, and Natalia N. Modjeska. 2003. Using the web for nominal anaphora resolution. In Proceedings of the EACL Workshop on the Computational Treatment of Anaphora. Budapest, Hungary, 14 April 2003, pages 39–46. find
  15. Katja Markert, Yufang Hou, and Michael Strube. 2012. Collective classification for fine-grained information status. In Proceedings of the 50th Annual Meeting of the Association for Computational Linguistics, Jeju Island, Korea, 8–14 July 2012, pages 795–804. find our card
  16. Ivan Meza-Ruiz and Sebastian Riedel. 2009. Jointly identifying predicates, arguments and senses using Markov logic. In Proceedings of Human Language Technologies 2009: The Conference of the North American Chapter of the Association for Computational Linguistics, Boulder, Col., 31 May – 5 June 2009, pages 155–163. find
  17. Shachar Mirkin, Ido Dagan, and Sebastian Padó. 2010. Assessing the role of discourse references in entailment inference. In Proceedings of the 48th Annual Meeting of the Association for Computational Linguistics, Uppsala, Sweden, 11–16 July 2010, pages 1209– 1219. find
  18. Natalia M. Modjeska, Katja Markert, and Malvina Nissim. 2003. Using the web in machine learning for other-anaphora resolution. In Proceedings of the 2003 Conference on Empirical Methods in Natural Language Processing, Sapporo, Japan, 11–12 July 2003, pages 176–183. find
  19. Malvina Nissim, Shipara Dingare, Jean Carletta, and Mark Steedman. 2004. An annotation scheme for information status in dialogue. In Proceedings of the 4th International Conference on Language Resources and Evaluation, Lisbon, Portugal, 26–28 May 2004, pages 1023–1026. find
  20. Robert Parker, David Graff, Junbo Kong, Ke Chen, and Kazuaki Maeda. 2011. English Gigaword Fifth Edition. LDC2011T07. find
  21. Massimo Poesio and Renata Vieira. 1998. A corpus-based investigation of definite description use. Computational Linguistics, 24(2):183–216. 2011. find
  22. Massimo Poesio, Rahul Mehta, Axel Maroudas, and Janet Hitzeman. 2004. Learning to resolve bridging references. In Proceedings of the 42nd Annual Meeting of the Association for Computational Linguistics, Barcelona, Spain, 21–26 July 2004, pages 143–150. find
  23. Massimo Poesio. 2003. Associate descriptions and salience: A preliminary investigation. In Proceedings of the EACL Workshop on the Computational Treatment of Anaphora. Budapest, Hungary, 14 April 2003, pages 31–38. find
  24. Hoifung Poon and Pedro Domingos. 2008. Joint unsupervised coreference resolution with Markov Logic. In Proceedings of the 2008 Conference on Empirical Methods in Natural Language Processing, Waikiki, Honolulu, Hawaii, 25–27 October 2008, pages 650– 659. find
  25. Ellen F. Prince. 1981. Towards a taxonomy of given-new information. In P. Cole, editor, Radical Pragmatics, pages 223–255. Academic Press, New York, N.Y. find
  26. Altaf Rahman and Vincent Ng. 2012. Learning the fine-grained information status of discourse entities. In Proceedings of the 13th Conference of the European Chapter of the Association for Computational Linguistics, Avignon, France, 23–27 April 2012, pages 798– 807. find
  27. Sebastian Riedel. 2008. Improving the accuracy and efficiency of MAP inference for Markov logic. In Proceedings of the 24th Conference on Uncertainty in Artificial Intelligence, Helsinki, Finland, 9–12 July 2008, pages 468–475. find
  28. Arndt Riester, David Lorenz, and Nina Seemann. 2010. A recursive annotation scheme for referential information status. In Proceedings of the 7th International Conference on Language Resources and Evaluation, La Valetta, Malta, 17–23 May 2010, pages 717–722. find
  29. Josef Ruppenhofer, Caroline Sporleder, Roser Morante, Collin Baker, and Martha Palmer. 2010. SemEval-2010 Task 10: Linking events and their participants in discourse. In Proceedings of the 5th International Workshop on Semantic Evaluations (SemEval-2), Uppsala, Sweden, 15–16 July 2010, pages 45–50. find
  30. Carina Silberer and Anette Frank. 2012. Casting implicit role linking as an anaphora resolution task. In Proceedings of STARSEM 2012: The First Joint Conference on Lexical and Computational Semantics, Montréal, Québec, Canada, 7–8 June 2012, pages 1– 10. find
  31. Wee Meng Soon, Hwee Tou Ng, and Daniel Chung Yong Lim. 2001. A machine learning approach to coreference resolution of noun phrases. Computational Linguistics, 27(4):521–544. find our card
  32. Renata Vieira and Massimo Poesio. 2000. An empirically-based system for processing definite descriptions. Computational Linguistics, 26(4):539– 593. find
  33. Ralph Weischedel, Martha Palmer, Mitchell Marcus, Eduard Hovy, Sameer Pradhan, Lance Ramshaw, Nianwen Xue, Ann Taylor, Jeff Kaufman, Michelle Franchini, Mohammed El-Bachouti, Robert Belvin, and Ann Houston. 2011. OntoNotes release 4.0. LDC2011T03, Philadelphia, Penn.: Linguistic Data Consortium. find
  34. Ian H. Witten and Eibe Frank. 2005. Data Mining: Practical Machine Learning Tools and Techniques. Morgan Kaufmann, San Francisco, Cal., 2nd edition. find