<?xml version="1.0" standalone="yes"?>
<Paper uid="I05-5012">
  <Title>a1 Information and Communication Technologies</Title>
  <Section position="7" start_page="94" end_page="94" type="concl">
    <SectionTitle>
6 Conclusion and Future Work
</SectionTitle>
    <Paragraph position="0"> In this paper, we presented an extension to the Viterbi algorithm which selects words in the string that are likely result in probable dependency structures. In a preliminary evaluation using precision and recall of dependency relations, we find that it improves grammaticality over a bigram model. In future work, we intend re-introduce the emission probabilities to model content selection. We also intend to use corpus-based dependency relation statistics and we would like to compare the two language models using perplexity. Finally, we would like to compare our system to that described in (Barzilay et al., 1999).</Paragraph>
  </Section>
class="xml-element"></Paper>