Susan Windisch Brown, Cecily Jill Duffield, Jena D. Hwang, Dmitrity Dligach, Sarah E. Vieweg, Jenny Davis, Martha Palmer - The Role of Functional Nouns in Manual Grouping of Verb Senses

Word sense ambiguity poses significant obstacles to accurate and efficient information extraction and automatic translation. Successful disambiguation of polysemous words in NLP applications depends on determining an appropriate level of granularity of sense distinctions, perhaps more so for distinguishing between multiple senses of verbs than for any other grammatical category. In this paper, we argue that clustering WordNet senses into more coarse-grained groupings based on the nominal arguments of the verb results in higher inter-annotator agreement and increased system performance. Identification of functional nouns are particularly useful in creating clear descriptions of verb sense clusters.
   Various criteria are considered when disambiguating senses and creating sense groupings for the verbs, including frequent lexical usages and collocations, syntactic features and alternations, and semantic features, similarly to Senseval2 (Palmer, et. al. 2006). While initial clustering is based on intuitions of the most salient categories, verb sense descriptions that do not contain information about nominal arguments and instead focus on semantic features of the verb often result in annotator confusion and low inter-annotator agreement scores. We have found features of nominal arguments, including descriptions of functional nouns, to be not only highly salient factors in creating verb sense clusters, but also crucial in producing clear verb sense descriptions that aid annotators in making consistent tagging choices.