<?xml version="1.0" standalone="yes"?>
<Paper uid="P05-1006">
  <Title>The Role of Semantic Roles in Disambiguating Verb Senses</Title>
  <Section position="3" start_page="0" end_page="42" type="intro">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0"> A word can have different meanings depending on the context in which it is used. Word Sense Disambiguation (WSD) is the task of determining the correct meaning (&amp;quot;sense&amp;quot;) of a word in context, and several efforts have been made to develop automatic WSD systems. Early work on WSD (Yarowsky, 1995) was successful for easily distinguishable homonyms like bank, which have multiple unrelated meanings. While homonyms are fairly tractable, highly polysemous verbs, which have related but subtly distinct senses, pose the greatest challenge for WSD systems (Palmer et al., 2001).</Paragraph>
    <Paragraph position="1"> Verbs are syntactically complex, and their syntax is thought to be determined by their underlying semantics (Grimshaw, 1990; Levin, 1993). Levin verb classes, for example, are based on the ability of a verb to occur in pairs of syntactic frames (diathesis alternations); different senses of a verb belong to different verb classes, which have different sets of syntactic frames that are supposed to reflect underlying semantic components that constrain allowable arguments. If this is true, then the correct sense of a verb should be revealed (at least partially) in its arguments.</Paragraph>
    <Paragraph position="2"> In this paper we show that the performance of automatic WSD systems can be improved by using richer linguistic features that capture information about predicate arguments and their semantic classes. We describe our approach to automatic WSD of verbs using maximum entropy models to combine information from lexical collocations, syntax, and semantic class constraints on verb arguments. The system performs at the best published accuracy on the English verbs of the Senseval-2 (Palmer et al., 2001) exercise on evaluating automatic WSD systems. The Senseval-2 verb instances have been manually tagged with their Word-Net sense and come primarily from the Penn Tree-bank WSJ. The WSJ corpus has also been manually annotated for predicate arguments as part of Prop-Bank (Kingsbury and Palmer, 2002), and the intersection of PropBank and Senseval-2 forms a corpus containing gold-standard annotations of WordNet senses and PropBank semantic role labels. This provides a unique opportunity to investigate the role of predicate arguments in verb sense disambiguation.</Paragraph>
    <Paragraph position="3"> We show that our system's accuracy improves significantly by adding features from PropBank, which explicitly encodes the predicate-argument informa- null tion that our original set of syntactic and semantic class features attempted to capture.</Paragraph>
  </Section>
class="xml-element"></Paper>