Elamur Research⌕
☰
home / papers / 2010 / Towards a Domain Independent Semantics: Enhancing Semantic Representation with Construction Grammar

Towards a Domain Independent Semantics: Enhancing Semantic Representation with Construction Grammar

Summary

The authors test whether ideas from Construction Grammar (CxG) can help improve the understanding of unfamiliar predicates. In CxG, meaning is carried not only by individual words but also by entire syntactic patterns. Classical WSD and SRL associate meaning primarily with lexemes. As a result, they handle rare verbs and texts from new domains poorly.

The study focuses on the caused-motion construction. In it, the subject causes the object to move along a path specified by a prepositional phrase, as in the sentence "someone sneezed the napkin off the table". From the WSJ portion of Penn Treebank II, the authors selected sentences with the structure (NP-SBJ (V NP PP)). For each one, they manually annotated whether it is genuinely caused-motion. On these data, they trained a classifier with features based on the preposition, the verb, VerbNet classes, function tags, and named entities.

The construction could be recognized with high accuracy. The preposition proved to be the most useful feature. Performance did not degrade on verbs that were absent from the training data. Thus, construction-level analysis can make semantics more robust to new vocabulary and domain shift.

Key points

  • About 14.7 thousand sentences matching the pattern (NP-SBJ (V NP PP)) were found in the WSJ. For annotation, the sample was reduced to approximately 1800 examples. In particular, cases with traces in the tree were removed: passives, questions, relative clauses.
  • Feature ablation analysis showed that removing the preposition causes the largest drop in performance: on the subset with OOV verbs, F falls to 0.69. The verb as a feature also has a statistically significant effect. Interactions between features are also noticeable, in particular between NEs for the object NP and VerbNet classes.
  • The test set contained 450 constructions. 92 of them (more than a fifth) contained verbs not seen in training — 83 distinct types. Accuracy and F-score on these examples did not drop: the difference is statistically insignificant (t=1.13, p>0.2).
  • For comparison, the authors mention CoNLL-2005: there, SRL systems degraded noticeably when moving from WSJ to Brown, mainly due to unfamiliar predicates.
  • Future plans: other construction types, more annotated data, and semantic features from FrameNet via SemLink.

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

Cite

Cite as Jena D. Hwang, Rodney D. Nielsen and Martha Palmer. 2010. Towards a Domain Independent Semantics: Enhancing Semantic Representation with Construction Grammar. Proceedings of the NAACL HLT Workshop on Extracting and Using Constructions in Computational Linguistics.

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

@inproceedings{hwang2010domain,
  title = {{Towards a Domain Independent Semantics: Enhancing Semantic Representation with Construction Grammar}},
  author = {Jena D. Hwang, Rodney D. Nielsen and Martha Palmer},
  booktitle = {Proceedings of the NAACL HLT Workshop on Extracting and Using Constructions in Computational Linguistics},
  year = {2010},
  url = {https://aclanthology.org/W10-0801/}
}

References

20 references, 5 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Agirre, Eneko and Philip Edmonds. 2006. Introduction. In Word Sense Disambiguation: Algorithms and Applications, Agirre and Edmonds (eds.), Springer. find
  2. Ben-David, Shai, Blitzer, John, Crammer, Koby Pereira, Fernando. 2006. 'Analysis of representations for domain adaptation', in NIPS. find
  3. Bikel, D., Schwartz, R., Weischedel, R. 1999. An algorithm that learns what’s in a name. Machine Learning: Special Issue on NL Learning, 34, 1-3. find
  4. Carreras, Xavier and Lluis Marquez. 2005. Introduction to the CoNLL- 2005 shared task: Semantic role labeling. Procs of CoNLL- 2005. our card
  5. Chih-Chung Chang and Chih-Jen Lin, LIBSVM : a library for support vector machines, 2001. Software available at http://www.csie.ntu.edu.tw/~cjlin/libsvm ↗
  6. Gildea, Daniel and Daniel Jurafsky. 2002. Automatic Labeling of Semantic Roles. Computational Linguistics 28:3, 245-288. our card
  7. Fillmore, Charles J., Christopher R. Johnson and Miriam R.L. Petruck (2003) Background to Framenet, International Journal of Lexicography, Vol 16.3: 235-250. find
  8. Fillmore, Charles, Paul Kay and Catherine O'Connor (1988). Regularity and Idiomaticity in Grammatical Constructions: The Case of let alone. Language 64: 501-38. find
  9. Giuglea, Ana-Maria and Alessandro Moschitti. 2006. Shallow semantic parsing based on FrameNet, Verb-Net and PropBank. In Proceedings of the 17th European Conference on Artificial Intelligence, Riva del Garda, Italy. find
  10. Goldberg, Adele E. 2006. Constructions at work. The nature of generalization in language. find
  11. Oxford: Oxford University Press Goldberg, Adele. E. 1995. Constructions: A construction grammar approach to argument structure. Chicago: University of Chicago Press. find
  12. Hovy, Edward H., Mitch Marcus, Martha Palmer, Sameer Pradhan, Lance Ramshaw, and Ralph M. Weischedel. 2006. OntoNotes: The 90% Solution. Short paper. Proceedings of the Human Language Technology / North American Association of Computational Linguistics conference (HLT-NAACL 2006). pp. 57-60, New Yor… find
  13. Kay, Paul. 2002. English Subjectless Tag Sentences. Language 78: 453-81. find
  14. Kipper-Schuler, Karin. 2005. VerbNet: A broad coverage, comprehensive verb lexicon. Ph.D. thesis, University of Pennsylvania. find
  15. Levin, Beth. 1993. English Verb Classes and Alternations: A Preliminary Investigation, University of Chicago Press, Chicago, IL. find
  16. Michaelis, Laura A. (2004). Type Shifting in Construction Grammar: An Integrated Approach to Aspectual Coercion. Cognitive Linguistics 15: 1-67. find
  17. Ng, Hwee Tou and Hian Beng Lee. 1996. Integrating multiple knowledge sources to disambiguate word sense: An exemplar-based approach. Proceedings of the 34th Annual Meeting of the Association for Computational Linguistics, Santa Cruz, California, 40–47. our card
  18. Marcus, Mitchell P, Santorini, Beatrice, Marcinkiewicz, Mary A. (1994) "Building a large annotated corpus of English: the Penn Treebank" Computational Linguistics 19: 313-330. our card
  19. Palmer, Martha. "Semlink: Linking PropBank, VerbNet and FrameNet." Proceedings of the Generative Lexicon Conference. Sept. 2009, Pisa, Italy: GenLex-09, 2009. find
  20. Palmer, Martha, Daniel Gildea, and Paul Kingsbury. 2005. The Proposition Bank: An Annotated Corpus of Semantic Roles. Computational Linguistics, 31(1):71–106. our card

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/hwang-2010-domain-independent-semantics-enhancing-semantic/