Elamur Research⌕
☰
home / papers / 2013 / Interpreting Anaphoric Shell Nouns using Antecedents of Cataphoric Shell Nouns as Training Data

Interpreting Anaphoric Shell Nouns using Antecedents of Cataphoric Shell Nouns as Training Data

Summary

Abstract anaphors such as this fact or this issue refer to propositions, events, or other complex content. Resolving them matters for summarization, information extraction, and non-factoid QA, but almost no annotated training data exists. Kolhatkar, Zinsmeister and Hirst work around this by using cataphoric shell nouns (CSNs) in patterns like the fact that X or the issue is whether X. In these patterns, simple syntactic rules can reliably pick out the content the noun points to. They automatically extracted such antecedents from a large slice of the New York Times corpus and used them to train ranking models. The models were then applied to the harder anaphoric shell nouns (ASNs), and the ranked candidates were judged through crowdsourcing. Across six shell nouns, top-ranked precision ranged from 0.35 to 0.72, consistently above a preceding-sentence baseline. This shows that freely available "easy" examples carry useful knowledge for resolving "hard" ones. The approach also holds up across domains, although it cannot use the anaphor's own context.

Key points

  • Data: The CSN training corpus has 211,722 NYT sentences matching Schmid's (2000) lexico-grammatical patterns (N-that, N-be-to, N-wh, etc.). The ASN test corpus has 2,323 instances.
  • Shell nouns: Six nouns from four of Schmid's semantic classes were covered: fact and reason (factual), issue and decision (mental), question (linguistic), and possibility (modal).
  • Precision at rank 1:
    • reason: up to 0.72 (baseline 0.44)
    • fact and question: about 0.70
    • issue: 0.47
    • decision: 0.35 (baseline 0.21)
    • For issue and decision, the correct antecedent appeared in the top four candidates about 76–81% of the time.
  • Cross-domain test: Applied to Kolhatkar and Hirst's (2012) MEDLINE this issue data, the automatically trained models reached P@1 = 0.45. This is comparable to that earlier system, which was trained on manually labelled data.
  • Limitations and future work:
    • Training sizes varied widely, from about 43,000 examples for fact to about 3,000 for issue.
    • CSN antecedents miss some properties of ASN antecedents. For example, decision training examples were mostly infinitive phrases, while the ASN antecedents were mostly full sentences.
    • The models cannot use the anaphor's context to choose between competing candidates.
    • Planned extensions include combining CSN and ASN data and adding contextual features such as the right-frontier rule.

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

9 papers of our library cite it (of the 140673 whose bibliographies we read); 4 of them have a page here, 5 not yet.

Cite

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

@inproceedings{kolhatkar2013interpreting,
  title = {{Interpreting Anaphoric Shell Nouns using Antecedents of Cataphoric Shell Nouns as Training Data}},
  author = {Kolhatkar, Varada and Zinsmeister, Heike and Hirst, Graeme},
  booktitle = {Proceedings of the 2013 Conference on Empirical Methods in Natural Language Processing},
  year = {2013},
  doi = {10.18653/v1/d13-1030},
  url = {https://doi.org/10.18653/v1/d13-1030}
}

References

22 references, 6 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Nicholas Asher. 1993. Reference to Abstract Objects in Discourse. Kluwer Academic Publishers, Dordrecht, Netherlands. find
  2. Donna K. Byron. 2004. Resolving pronominal reference to abstract entities. Ph.D. thesis, Rochester, New York: University of Rochester. find
  3. Pascal Denis and Jason Baldridge. 2008. Specialized models and ranking for coreference resolution. In Proceedings of the 2008 Conference on Empirical Methods in Natural Language Processing, pages 660–669, Honolulu, Hawaii, October. Association for Computational Linguistics. find our card
  4. Miriam Eckert and Michael Strube. 2000. Dialogue acts, synchronizing units, and anaphora resolution. Journal of Semantics, 17:51–89. find our card
  5. Gill Francis. 1986. Anaphoric Nouns. Discourse Analysis Monographs 11, Birmingham: English Language Research, University of Birmingham. find
  6. M. A. K. Halliday and Ruqaiya Hasan. 1976. Cohesion in English. Longman Pub Group. find
  7. Marti A. Hearst. 1992. Automatic acquisition of hyponyms from large text corpora. In In Proceedings of the 14th International Conference on Computational Linguistics, pages 539–545, Nantes, France. Association for Computational Linguistics. find
  8. Roz Ivanic. 1991. Nouns in search of a context: A study of nouns with both open- and closed-system characteristics. International Review of Applied Linguistics in Language Teaching, 29:93–114. find
  9. Thorsten Joachims. 2002. Optimizing search engines using clickthrough data. In ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD), pages 133–142. find
  10. Varada Kolhatkar and Graeme Hirst. 2012. Resolving “this-issue” anaphora. In Proceedings of the 2012 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning, pages 1255–1265, Jeju Island, Korea, July. Association for Computational Linguistics. find
  11. Varada Kolhatkar, Heike Zinsmeister, and Graeme Hirst. 2013. Annotating anaphoric shell nouns with their antecedents. In Proceedings of the 7th Linguistic Annotation Workshop and Interoperability with Discourse, pages 112–121, Sofia, Bulgaria, August. Association for Computational Linguistics. find
  12. Daniel Marcu and Abdessamad Echihabi. 2002. An unsupervised approach to recognizing discourse relations. In Proceedings of the 40th Annual Meeting on Association for Computational Linguistics, pages 368–375, Stroudsburg, PA, USA. Association for Computational Linguistics. find our card
  13. Mitchell Marcus, Beatrice Santorini, and Mary Ann Marcinkiewicz. 1993. Building a large annotated corpus of english: The penn treebank. Computational Linguistics, 19(2):313–330. find our card
  14. Christoph Müller. 2008. Fully Automatic Resolution of It, This and That in Unrestricted Multi-Party Dialog. Ph.D. thesis, Universität Tübingen. find
  15. Rebecca J. Passonneau. 1989. Getting at discourse referents. In Proceedings of the 27th Annual Meeting of the Association for Computational Linguistics, pages 51– 59, Vancouver, British Columbia, Canada. Association for Computational Linguistics. find
  16. Massimo Poesio, Simone Ponzetto, and Yannick Versley. 2011. Computational models of anaphora resolution: A survey. Unpublished. find
  17. Hans-Jörg Schmid. 2000. English Abstract Nouns As Conceptual Shells: From Corpus to Cognition. Topics in English Linguistics 34. Mouton de Gruyter, Berlin. find
  18. Wee Meng Soon, Hwee Tou Ng, and Chung Yong Lim. 2001. A machine learning approach to coreference resolution of noun phrases. Computational Linguistics, 27(4):521–544. find our card
  19. Michael Strube and Christoph Müller. 2003. A machine learning approach to pronoun resolution in spoken dialogue. In Proceedings of the 41st Annual Meeting of the Association for Computational Linguistics, pages 168–175, Sapporo, Japan, July. Association for Computational Linguistics. find our card
  20. Zeno Vendler. 1968. Adjectives and Nominalizations. Mouton de Gruyter, The Hague. find
  21. Bonnie Lynn Webber. 1991. Structure and ostension in the interpretation of discourse deixis. In Language and Cognitive Processes, pages 107–135. find
  22. Yiming Yang and Jan O. Pedersen. 1997. A comparative study on feature selection in text categorization. In Proceedings of the 14th International Conference on Machine Learning, pages 412–420, Nashville, TN. find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/papers/kolhatkar-2013-interpreting-anaphoric-shell-nouns-antecedents/