<?xml version="1.0" standalone="yes"?>
<Paper uid="W05-0632">
  <Title>Maximum Entropy based Semantic Role Labeling</Title>
  <Section position="5" start_page="211" end_page="211" type="concl">
    <SectionTitle>
4 Conclusion
</SectionTitle>
    <Paragraph position="0"> We have presented a two-phase SRL method based on a ME model. In the two-phase method, in order to improve the performance of identification that dominate the overall performance, we have performed pre-processing. Experimental results show that our system obtains an Fb=1 of 72.68% on the WSJ test and that the introduction of pre-processing improves the performance, as compared with the case when all parse constituents are considered.</Paragraph>
  </Section>
class="xml-element"></Paper>