Elamur Research⌕
☰
home / papers / 2010 / Beyond NomBank: A Study of Implicit Arguments for Nominal Predicates

Beyond NomBank: A Study of Implicit Arguments for Nominal Predicates

Paper · 2010 · Matthew Gerber and Joyce Chai

Summary

Standard semantic role labeling looks for a predicate's arguments only inside its own sentence, and often only in its nearby syntax. NomBank therefore leaves many roles of nominal predicates empty, even when a reader can easily fill them from the same sentence or from earlier text. Gerber and Chai call these missing fillers implicit arguments. They annotated them for ten frequent nominal predicates (such as price, sale, investor, loss, bid and loan) across the training, development and test sections of the Penn TreeBank. For each empty role they searched the current sentence and all preceding ones. With these annotations, role coverage rises by 65% relative to NomBank alone. They then trained a supervised classifier that decides whether a candidate constituent, together with its whole coreference chain, fills a missing role. The classifier's features combine verbal and nominal SRL evidence. It beats an informed baseline by 15 F1 points (59% relative). The work shows that implicit arguments are common and can be recovered automatically. It sets a baseline for discourse-level role filling.

Key points

  • The annotation covers 1,253 instances of 10 predicates. Overall role coverage grew from 28.0% to 46.2%, a gain of 18 points (65% relative). An independent student annotator reached a Cohen's kappa of 67% against the authors.
  • The predicates were chosen by multiplying two factors: how many fewer roles the noun form fills in NomBank than the verb form fills in PropBank, and how frequent the noun is.
  • About 56% of implicit arguments are found in the predicate's own sentence. Some lie as far as 46 sentences back, but most are within the previous few sentences.
  • The model classifies a tuple ⟨predicate, iarg position, coreference chain⟩. It draws evidence from the other mentions of an entity and their SRL roles, and uses coreference patterns mined from Gigaword.
  • The candidate window was limited to two sentences, and only 80% of the implicit arguments are reachable inside it. Non-eventive collocations such as "mutual fund" and "investment banker" caused false positives.

Interesting ideas

  • I1. The whole coreference chain is the unit to classify, not a single mention. Every mention's role in other events then counts as evidence: an entity that "needs investment" elsewhere becomes a likely investee.
  • I2. Implicit roles are learned through cross-predicate tendencies: exporters and suppliers tend to also be sellers, and investors tend to sell what they invested in. This lets a role transfer from one event to a related one.
  • I3. Target selection is cheap and data-driven: compare how many roles a noun realizes with how many its verb realizes, then weight by frequency.
  • I4. The informed baseline is simple: an identical predicate nearby shares its argument in the same slot. The analysis focuses on cases where the model wins and this baseline fails.
  • I5. Non-eventive collocations ("stock fund") should not get implicit arguments at all. Filtering them protects precision.

For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

  • I1 (frames): when filling an empty role of a frame, score whole coreference chains and use each chain's roles in other frames as explainable support.
  • I2 (world): compile such predicate-to-predicate role links into the knowledge layer, so the world model can infer that the earlier exporter is the agent of a later "sale."
  • I3 (learn): use the noun-verb role gap to decide which nominal frames deserve silver annotation or LLM-teacher effort first.
  • I4 (learn): keep such a trivial baseline as a control in evaluation, so gains in implicit-role filling are not just repeats of the same predicate.
  • I5 (frames): before instantiating a nominal frame, check whether the noun is used in an eventive sense, and skip role filling for lexicalized compounds.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓Michigan State UniversityEast Lansing · United States23 works · 1,077 citations

 Fields, topics and concepts

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 works

🧭 Topics 1Research subjects and languages the work is about.Semantic role labeling · 145 works

Cite

Cite as Matthew Gerber and Joyce Chai. 2010. Beyond NomBank: A Study of Implicit Arguments for Nominal Predicates. Proceedings of the 48th Annual Meeting of the Association for Computational Linguistics.

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

@inproceedings{gerber2010beyond,
  title = {{Beyond NomBank: A Study of Implicit Arguments for Nominal Predicates}},
  author = {Matthew Gerber and Joyce Chai},
  booktitle = {Proceedings of the 48th Annual Meeting of the Association for Computational Linguistics},
  year = {2010},
  url = {https://aclanthology.org/P10-1160/}
}

References

26 references, 9 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Aljoscha Burchardt, Anette Frank, and Manfred Pinkal. 2005. Building text meaning representations from contextually related frames - a case study. In Proceedings of the Sixth International Workshop on Computational Semantics. find
  2. Xavier Carreras and Lluı́s Màrquez. 2005. Introduction to the CoNLL-2005 shared task: Semantic role labeling.
  3. Nathanael Chambers and Dan Jurafsky. 2008. Unsupervised learning of narrative event chains. In Proceedings of the Association for Computational Linguistics, pages 789–797, Columbus, Ohio, June. Association for Computational Linguistics. find
  4. Jacob Cohen. 1960. A coefficient of agreement for nominal scales. Educational and Psychological Measurement, 20(1):3746. ↗
  5. Bradley Efron and Robert J. Tibshirani. 1993. An Introduction to the Bootstrap. Chapman & Hall, New York. ↗
  6. 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. ↗
  7. Christiane Fellbaum. 1998. WordNet: An Electronic Lexical Database (Language, Speech, and Communication). The MIT Press, May. find
  8. C.J. Fillmore and C.F. Baker. 2001. Frame semantics for text understanding. In Proceedings of WordNet and Other Lexical Resources Workshop, NAACL. find
  9. Matthew Gerber, Joyce Y. Chai, and Adam Meyers. 2009. The role of implicit argumentation in nominal SRL. In Proceedings of the North American Chapter of the Association for Computational Linguistics, pages 146–154, Boulder, Colorado, USA, June. find
  10. David Graff. 2003. English Gigaword. Linguistic Data Consortium, Philadelphia. find
  11. Jan Hajič, Massimiliano Ciaramita, Richard Johansson, Daisuke Kawahara, Maria Antònia Martı́, Lluı́s Màrquez, Adam Meyers, Joakim Nivre, Sebastian Padó, Jan Štěpánek, Pavel Straňák, Mihai Surdeanu, Nianwen Xue, and Yi Zhang. 2009. The CoNLL-2009 shared task: Syntactic and semantic dependencies in multiple languages. In Proceedings of the Thirteenth Conference on Computational Natural Language Learning (CoNLL 2009): Shared Task, pages 1–18, Boulder, Colorado, June. Association for Computational Linguistics. ↗
  12. Ryu Iida, Mamoru Komachi, Kentaro Inui, and Yuji Matsumoto. 2007. Annotating a Japanese text corpus with predicate-argument and coreference relations. In Proceedings of the Linguistic Annotation Workshop in ACL-2007, page 132139.
  13. Kenji Imamura, Kuniko Saito, and Tomoko Izumi. 2009. Discriminative approach to predicate-argument structure analysis with zero-anaphora resolution. In Proceedings of the ACL-IJCNLP 2009 Conference Short Papers, pages 85–88, Suntec, Singapore, August. Association for Computational Linguistics.
  14. P. Kingsbury, M. Palmer, and M. Marcus. 2002. Adding semantic annotation to the Penn TreeBank. In Proceedings of the Human Language Technology Conference (HLT’02). find
  15. Karin Kipper. 2005. VerbNet: A broad-coverage, comprehensive verb lexicon. Ph.D. thesis, Department of Computer and Information Science University of Pennsylvania. find
  16. Mitchell Marcus, Beatrice Santorini, and Mary Ann Marcinkiewicz. 1993. Building a large annotated corpus of English: the Penn TreeBank. Computational Linguistics, 19:313–330. ↗
  17. Adam Meyers. 2007. Annotation guidelines for NomBank - noun argument structure for PropBank. Technical report, New York University. find
  18. Martha S. Palmer, Deborah A. Dahl, Rebecca J. Schiffman, Lynette Hirschman, Marcia Linebarger, and John Dowding. 1986. Recovering implicit information. In Proceedings of the 24th annual meeting on Association for Computational Linguistics, pages 10–19, Morristown, NJ, USA. Association for Computational Linguistics. find
  19. Patrick Pantel and Deepak Ravichandran. 2004. Automatically labeling semantic classes. In Daniel Marcu Susan Dumais and Salim Roukos, editors, HLT-NAACL 2004: Main Proceedings, pages 321–328, Boston, Massachusetts, USA, May 2 - May 7. Association for Computational Linguistics.
  20. Rashmi Prasad, Alan Lee, Nikhil Dinesh, Eleni Miltsakaki, Geraud Campion, Aravind Joshi, and Bonnie Webber. 2008. Penn discourse treebank version 2.0. Linguistic Data Consortium, February. ↗
  21. P. Pudil, J. Novovicova, and J. Kittler. 1994. Floating search methods in feature selection. Pattern Recognition Letters, 15:1119–1125. ↗
  22. Vasin Punyakanok, Dan Roth, and Wen-tau Yih. 2008. The importance of syntactic parsing and inference in semantic role labeling. Comput. Linguist., 34(2):257–287. ↗
  23. Josef Ruppenhofer, Caroline Sporleder, Roser Morante, Collin Baker, and Martha Palmer. 2009. Semeval-2010 task 10: Linking events and their participants in discourse. In Proceedings of the Workshop on Semantic Evaluations: Recent Achievements and Future Directions (SEW-2009), pages 106–111, Boulder, Colorado, June. Association for Computational Linguistics.
  24. Ryohei Sasano, Daisuke Kawahara, and Sadao Kurohashi. 2004. Automatic construction of nominal case frames and its application to indirect anaphora resolution. In Proceedings of Coling 2004, pages 1201–1207, Geneva, Switzerland, Aug 23–Aug 27. COLING. ↗
  25. Mihai Surdeanu, Richard Johansson, Adam Meyers, Lluı́s Màrquez, and Joakim Nivre. 2008. find
  26. The CoNLL 2008 shared task on joint parsing of syntactic and semantic dependencies. In CoNLL 2008: Proceedings of the Twelfth Conference on Computational Natural Language Learning, pages 159–177, Manchester, England, August. Coling 2008 Organizing Committee.

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/gerber-2010-beyond-nombank-study-implicit-arguments/