<?xml version="1.0" standalone="yes"?>
<Paper uid="W98-0702">
  <Title>I I I I I I I I I I I I I I I I I I I Disambiguating Verbs with the WordNet Category of the Direct Object</Title>
  <Section position="9" start_page="13" end_page="14" type="concl">
    <SectionTitle>
8 Conclusions and Future Work
</SectionTitle>
    <Paragraph position="0"> The semantic category of the direct object plays a major role in determining the aspectual class of a clause. To demonstrate this, a rule was developed that uses WordNet categories to classify have-clauses according to stativity. When evaluated over an unrestricted set of nouns, this rule achieved an accuracy of 79.6%, compared to the baseline performance of 69.9%. Moreover, a favorable tradeoff in recall was achieved, attaining 67.7% event recall, compared to the the baseline's 0.0%. More specifically, frequent WordNet categories were shown to predict aspectual class with an average precision of 82.7%. These results are impressive, considering the unresolved semantic ambiguity of direct objects, and the technical terminology of the medical domain.</Paragraph>
    <Paragraph position="1"> WordNet categories also improved the classification performance of linguistic indicators for completedness. Although more sparsely measured, the accuracy achieved by indicators measured over multiple constituents is comparable to that of indicators measured over the verb only, with a favorable trade- null off in recall. Therefore, the noise introduced by this more sparse measurement of indicators is more than compensated for by the ability to resolve aspectually ambiguous verbs.</Paragraph>
    <Paragraph position="2"> Furthermore, I have derived a semantic hierarchy of statively ambiguous verbs in order to predict verbs' subcategorization frames. This in turn guides the disambiguation of such verbs. Future work will investigate whether rules such as that developed for have could apply over multiple verbs that share sub-categorization behavior. Additionally, it is possible that WordNet's categorization of verbs could automatically place verbs into these semantic groups.</Paragraph>
    <Paragraph position="3"> Finally, disambiguating the direct object according to WordNet categories, e.g., Resnik (1995), would improve the accuracy of using these categories to disambiguate verbs.</Paragraph>
  </Section>
class="xml-element"></Paper>