Elamur Research⌕
☰
home / papers / 2016 / Sense Anaphoric Pronouns: Am I One?

Sense Anaphoric Pronouns: Am I One?

Summary

The paper addresses sense anaphora, in which a pronoun takes its meaning from an earlier expression but refers to a different entity, as in "existing ones" pointing back to "loans". Earlier computational work studied only one. This study covers a much broader set of forms: one(s), indefinite quantifiers, numerals, demonstratives (that, those) and other/another. The authors wrote annotation guidelines and had four annotators mark anaphors and their antecedents in a third of English OntoNotes. The result is the SAnaNotes corpus, which they release publicly. They then trained a LIBLINEAR SVM that separates anaphoric uses from other uses, such as generics and numerals. The classifier uses only local lexical and syntactic features. On one/ones it clearly outperforms reimplementations of Gardiner (2003) and Ng et al. (2005), and it also extends to the new anaphor categories. Numerals and indefinites remain difficult because there are few positive examples. The work matters because unresolved sense anaphors leave a gap in what a text means. The authors expect such anaphors to be more frequent in conversational input, such as voice queries.

Key points

  • SAnaNotes covers 1 163 OntoNotes documents, sampled to preserve the domain mix. It is released at github.com/dmorr-google/sense-anaphora.
  • Inter-annotator agreement was measured on 25 documents annotated by all four annotators. Anaphor identification reached Fleiss' κ = .67, and antecedent span agreement was 63%. The phenomenon is rare, at about 0.6 anaphors per document and 4.3% / 3.3% anaphoric candidates in the train / test splits.
  • The features cover the candidate token, a two-token window on each side, the syntactic head and the leftmost child. For each of these, the model uses word, POS, dependency label and Brown cluster, plus conjoined pairs, giving 31 feature types in total.
  • On one/ones, the system reaches 79.01% F1. The reimplemented Gardiner (2003) system scores 56.20% and Ng et al. (2005) scores 58.80%.
  • Across all categories, the system reaches 61.80% F1, or 67.34% excluding numerals. Ng et al.'s features score 31.30% and 38.60% on the same settings.

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

5 papers of our library cite it (of the 140673 whose bibliographies we read); 2 of them have a page here, 3 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/w16-0701

@inproceedings{recasens2016sense,
  title = {{Sense Anaphoric Pronouns: Am I One?}},
  author = {Recasens, Marta and Hu, Zhichao and Rhinehart, Olivia},
  booktitle = {Proceedings of the Workshop on Coreference Resolution Beyond OntoNotes (CORBON 2016)},
  year = {2016},
  doi = {10.18653/v1/w16-0701},
  url = {https://doi.org/10.18653/v1/w16-0701}
}

References

20 references, 3 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Joan Bresnan. 1971. A note on the notion “identity of sense anaphora”. Linguistic Inquiry, 2:589–597. Peter F. Brown, Peter V. deSouza, Robert L. Mercer, Vincent J. Della Pietra, and Jenifer C. Lai. 1992. Class-based n-gram models of natural language. Computational Linguistics, 18(4):467–479. find
  2. Herbert H. Clark. 1975. Bridging. In R. C. Schank and B. L. Nash-Webber, editors, Theoretical Issues in Natural Language Processing, pages 169–174. find
  3. Deborah Anna Dahl. 1985. The structure and function of one-anaphora in English. Ph.D. thesis, University of Minnesota. find
  4. Mary Dalrymple, Stuart M. Shieber, and Fernando C.N. Pereira. 1991. Ellipsis and higher order unification. Linguistics and Philosophy, 14:399–452. find
  5. 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
  6. Joseph L. Fleiss. 1971. Measuring nominal scale agreement among many raters. Psychological Bulletin, 76(5):378–382. find
  7. Mary Gardiner. 2003. Identifying and resolving one-anaphora. B.S. Thesis, Department of Computing, Macquarie University. find
  8. John Grinder and Paul M. Postal. 1971. Missing antecedents. Linguistic Inquiry, 2:589–597. find
  9. Mark Hall, Eibe Frank, Geoffrey Holmes, Bernhard Pfahringer, Peter Reutemann, and Ian H. Witten. 2009. The WEKA data mining software: An update. SIGKDD Explorations, 11:10–18. find
  10. Grame Hirst. 1981. Anaphora in natural language understanding: a survey. Lecture Notes in Computer Science. find
  11. Springer-Verlag, Berlin, New York. J. Richard Landis and Gary G. Koch. 1977. The measurement of observer agreement for categorical data. Biometrics, 33(1):159–174. find
  12. Susann Luperfoy. 1991. Discourse Pegs: A Computational Analysis of Context-Dependent Referring Expressions. Ph.D. thesis, University of Texas at Austin. find
  13. Katja Markert and Malvina Nissim. 2005. Comparing knowledge sources for nominal anaphora resolution. Computational Linguistics, 31:367–401. find our card
  14. Katja Markert, Yufang Hou, and Michael Strube. 2012. Collective classification for fine-grained information status. In Proceedings of ACL, pages 795–804. find our card
  15. Hwee Tou Ng, Yu Zhou, Robert Dale, and Mary Gardiner. 2005. A machine learning approach to identification and resolution of one-anaphora. In Proceedings of IJCAI, pages 1105–1110. find
  16. Joakim Nivre, Johan Hall, Jens Nilsson, Atanas Chanev, Gülşen Eryiǧit, Sandra Kübler, Svetoslav Marinov, and Erwin Marsi. 2007. MaltParser: A language-independent system for data-driven dependency parsing. Natural Language Engineering, 13:95–135. find
  17. John Payne, Geoffrey K. Pullum, Barbara C. Scholz, and Eva Berlage. 2013. Anaphoric one and its implications. Language, 89(4):794–829. find
  18. Massimo Poesio, Florence Bruneseaux, and Laurent Romary. 1999. The MATE meta-scheme for coreference in dialogues in multiple languages. In Proceedings of the ACL Workshop Towards Standards and Tools for Discourse Tagging, pages 65–74. find our card
  19. Ivan A. Sag. 1976. Deletion and logical form. Ph.D. thesis, Massachusetts Institute of Technology. find
  20. Ralph Weischedel, Eduard Hovy, Mitchell Marcus, Martha Palmer, Robert Belvin, Sameer Pradhan, Lance Ramshaw, and Nianwen Xue. 2011. OntoNotes: A large training corpus for enhanced processing. In Joseph Olive, Caitlin Christianson, and John McCary, editors, Handbook of Natural Language Processing and… find

↑ back to top

This page in Markdown →