<?xml version="1.0" standalone="yes"?>
<Paper uid="P06-1033">
  <Title>Graph Transformations in Data-Driven Dependency Parsing</Title>
  <Section position="8" start_page="262" end_page="262" type="concl">
    <SectionTitle>
5 Conclusion
</SectionTitle>
    <Paragraph position="0"> The results presented in this paper confirm that choosing the right representation is important in parsing. By systematically transforming the representation of coordinate structures and verb groups in PDT, we achieve a 10% error reduction for a data-driven dependency parser. Adding graph transformations for non-projective dependency parsing gives a total error reduction of about 20% (even more for unlabeled exact match).</Paragraph>
    <Paragraph position="1"> In this way, we achieve state-of-the-art accuracy with a deterministic, classifier-based dependency parser.</Paragraph>
  </Section>
class="xml-element"></Paper>