<?xml version="1.0" standalone="yes"?>
<Paper uid="W06-2938">
  <Title>Dependency Parsing as a Classi cation Problem</Title>
  <Section position="8" start_page="248" end_page="249" type="concl">
    <SectionTitle>
5 Conclusion
</SectionTitle>
    <Paragraph position="0"> I used standard machine learning techniques to investigate the lower bound accuracy and the impact of various attributes on the subproblem of identifying dependency links between adjacent words. The technique was then extended to identify long distance dependencies and used as a parser. The model gives average results for Turkish and Japanese but 2This counterintuitive procedure was used because it gave the best results on the training set.</Paragraph>
    <Paragraph position="1">  generally performs below average. The lack of a specialized parsing algorithm taking into account sentence wide constraints and the lack of a probabilistic component in the model are probably to blame. Nevertheless, the particular decomposition of the problem and the simplicity of the resulting models provide some insight into the dif culties associated with individual languages.</Paragraph>
  </Section>
class="xml-element"></Paper>