<?xml version="1.0" standalone="yes"?>
<Paper uid="W05-1506">
  <Title>Better k-best Parsing</Title>
  <Section position="10" start_page="59" end_page="59" type="concl">
    <SectionTitle>
6 Conclusion
</SectionTitle>
    <Paragraph position="0"> The problem of k-best parsing and the effect of k-best list size and quality on applications are subjects of increasing interest for NLP research. We have presented here a general-purpose algorithm for k-best parsing and applied it to two state-of-the-art, large-scale NLP systems: Bikel's implementation of Collins' lexicalized PCFG model (Bikel, 2004; Collins, 2003) and Chiang's synchronous CFG based decoder (Chiang, 2005) for machine translation. We hope that this work will encourage further investigation into whether larger and better k-best lists will improve performance in NLP applications, questions which we ourselves intend to pursue as well.</Paragraph>
  </Section>
class="xml-element"></Paper>