<?xml version="1.0" standalone="yes"?>
<Paper uid="W05-1628">
  <Title>2Information and Communication Technologies</Title>
  <Section position="8" start_page="0" end_page="0" type="concl">
    <SectionTitle>
7 Conclusion and Future Work
</SectionTitle>
    <Paragraph position="0"> In this paper, we presented an extension to the Viterbi algorithm that statistically determines dependency structure of partially generated sentences and selects of words that are likely to attach to this structure. The resulting sentence is more grammatical than that generated using a bigram baseline. In future work, we intend to conduct experiments to see whether the smoothing approaches chosen are successful in parsing without introducing spurious dependency relations.</Paragraph>
    <Paragraph position="1"> We would also like to re-integrate the emission probability (that is, the word content selection model). We are also in the process of developing a measure of consistency. Finally, we intend to provide a comparison evaluation with Barzilay's Information Fusion work.</Paragraph>
  </Section>
class="xml-element"></Paper>