<?xml version="1.0" standalone="yes"?>
<Paper uid="W04-2116">
  <Title>Empirical Acquisition of Differentiating Relations from Definitions</Title>
  <Section position="6" start_page="0" end_page="0" type="concl">
    <SectionTitle>
5 Related work
</SectionTitle>
    <Paragraph position="0"> Most of the work addressing differentia extraction has relied upon manually constructed extraction rules (Vanderwende, 1996; Barri`ere, 1997; Rus, 2002). Here the emphasis is switched from transformation patterns for extracting relations into statistical classification for relation disambiguation, given tagged corpora with examples. This allows for better coverage at the expense of precision. Note that relation disambiguation is not yet addressed in Extended WordNet (Rus, 2002); for example, prepositions are treated as predicates in the logical form representation. Their extraction process is also closely tied into the specifics of the parser, as a transformation rule is developed for each grammar rule.</Paragraph>
    <Paragraph position="1"> This work addresses the acquisition of conceptual distinctions. In principle, it can handle any level of granularity given sufficient training data; however, addressing distinctions at the level of near-synonyms (Edmonds and Hirst, 2002) might require customized analysis for each cluster of nearly synonymous words. Inkpen and Hirst (2001) discuss how this can be automated by analyzing specialized synonymy dictionaries. Decision lists of indicative keywords are learned for the broad types of pragmatic distinctions, and these are then manually split into decision lists for more-specific distinctions.</Paragraph>
    <Section position="1" start_page="0" end_page="0" type="sub_section">
      <SectionTitle>
6Conclusion
</SectionTitle>
      <Paragraph position="0"> We have presented an empirical methodology for extracting information from dictionary definitions.</Paragraph>
      <Paragraph position="1"> This differs from previous approaches by using data-driven relation disambiguation, using FrameNet semantic roles annotations mapped into a reduced inventory. All the definitions from WordNet 1.7.1 were analyzed using this process, and the results evaluated by four human judges. The overall results were not high, but the evaluation was comparable to relations that were manually corrected before coding.</Paragraph>
    </Section>
  </Section>
class="xml-element"></Paper>