<?xml version="1.0" standalone="yes"?>
<Paper uid="N04-1030">
  <Title>Shallow Semantic Parsing using Support Vector Machines</Title>
  <Section position="13" start_page="0" end_page="0" type="concl">
    <SectionTitle>
15 Conclusions
</SectionTitle>
    <Paragraph position="0"> We have described an algorithm which significantly improves the state-of-the-art in shallow semantic parsing.</Paragraph>
    <Paragraph position="1"> Like previous work, our parser is based on a supervised machine learning approach. Key aspects of our results include significant improvement via an SVM classifier, improvement from new features and a series of analytic experiments on the contributions of the features. Adding features that are generalizations of the more specific features seemed to help. These features were named entities, head word part of speech and verb clusters. We also analyzed the transferability of the features to a new text source.</Paragraph>
    <Paragraph position="2"> We would like to thank Ralph Weischedel and Scott Miller of BBN Inc. for letting us use their named entity tagger - IdentiFinder; Martha Palmer for providing us with the PropBank data, Valerie Krugler for tagging the AQUAINT test set with PropBank arguments, and all the anonymous reviewers for their helpful comments.</Paragraph>
  </Section>
class="xml-element"></Paper>