<?xml version="1.0" standalone="yes"?>
<Paper uid="P06-1113">
  <Title>Question Answering with Lexical Chains Propagating Verb Arguments</Title>
  <Section position="8" start_page="902" end_page="903" type="concl">
    <SectionTitle>
5 Conclusion
</SectionTitle>
    <Paragraph position="0"> This paper describes the approach of propagating verb arguments along lexical chains with Word-Net relations using VerbNet frames. Since VerbNet frames are not associated with all verb senses from WordNet, some verb senses were added automatically to the existing VerbNet frames. The algorithm was used to improve the performance of the answer's ranking stage in Question Answering system. Only a restricted set of WordNet semantic  Arguments propagated Top 5 chains Top 10 chains Top 20 chains a all arguments 23(21.6%) 28(26.4%) 32(30.2%) b at least one argument 73(68.8%)% 81(76.4%) 89(83.9%) c no arguments 32(30.2%) 25(23.6%) 17(16.0%) relations were used to propagate predicate arguments. Lexical chains were also derived between the arguments for a better match. On the set of factoid questions from TREC 2004, it was found that for 33(14.3%) questions, the words in the question and the related words in the answer could be linked using lexical chains containing only the relations from the restricted set that propagate verb arguments. Overall, the algorithm for propagating verb arguments improved the system performance with 2.4%</Paragraph>
  </Section>
class="xml-element"></Paper>