<?xml version="1.0" standalone="yes"?>
<Paper uid="C04-1204">
  <Title>Deep Linguistic Analysis for the Accurate Identification of Predicate-Argument Relations</Title>
  <Section position="7" start_page="2" end_page="2" type="concl">
    <SectionTitle>
6 Conclusion
</SectionTitle>
    <Paragraph position="0"> In this paper, the accuracy of HPSG parsing was evaluated in terms of the identification of predicate-argument relations. By assuming unique mapping from HPSG predicate argument structures into the PropBank annotation of semantic arguments, we could directly compare the output of an HPSG parser with PropBank. Despite not using Prop-Bank for the training of a disambiguation model, the HPSG parser achieved a high accuracy competitive with the previous studies on the identification of PropBank annotations. This result reveals the accurate identification of predicate-argument relations by HPSG parsing.</Paragraph>
    <Paragraph position="1"> Although this study directly compared the HPSG output with PropBank, we may require an additional machine learning step as in the existing studies to obtain higher accuracy because the accuracy attained by gold parses showed a limitation of our approach. Another possibility is to directly extract PropBank-style semantic representations by reforming the grammar extraction algorithm (Chen and Rambow, 2003), and to estimate a disambiguation model using the PropBank.</Paragraph>
  </Section>
class="xml-element"></Paper>