<?xml version="1.0" standalone="yes"?>
<Paper uid="N03-1007">
  <Title>Performance QA System Using Lexico-Semantic Pattern Matching and Shallow NLP&amp;quot;, Proceedings of</Title>
  <Section position="8" start_page="0" end_page="0" type="concl">
    <SectionTitle>
9 Conclusion
</SectionTitle>
    <Paragraph position="0"> It has been shown that recognizing that a clarification dialogue is occurring can simplify the task of retrieving an answer by constraining the subset of documents in which an answer is to be found. An algorithm was presented to recognize the occurrence of clarification dialogue and is shown to have a good performance. The major limitation of our algorithm is the fact that it only considers series of questions, not series of answers. As noted above, it is often necessary to look at an answer to a question to determine whether the current question is a clarification question or not. Our sentence similarity algorithm was limited by the number of semantic relationships in WordNet: for example, a big improvement would come from the use of noun-verb relationships. Future work will be directed on extending WordNet in this direction and in providing other useful semantic relationships. Work also needs to be done on using information given by answers, not just questions in recognizing clarification dialogue and on coping with the cases in which clarification dialogue recognition is not enough to retrieve an answer and where other, more complex, techniques need to be used. It would also be beneficial to examine the use of a similarity function in which similarity decayed in function of the distance in time between the current question and the past questions.</Paragraph>
  </Section>
class="xml-element"></Paper>