<?xml version="1.0" standalone="yes"?>
<Paper uid="E06-1050">
  <Title>A Probabilistic Answer Type Model</Title>
  <Section position="7" start_page="399" end_page="399" type="concl">
    <SectionTitle>
6 Conclusions
</SectionTitle>
    <Paragraph position="0"> We have presented an unsupervised probabilistic answer type model. Our model uses contexts derived from the question and the candidate answer to calculate the appropriateness of a candidate answer. Statistics gathered from a large corpus of text are used in the calculation, and the model is constructed to exploit these statistics without being overly specific or overly general.</Paragraph>
    <Paragraph position="1"> The method presented here avoids the use of an explicit list of answer types. Explicit answer types can exhibit poor performance, especially for those questions not fitting one of the types. They must alsoberedefinedwheneitherthedomainorcorpus substantially changes. By avoiding their use, our answer typing method may be easier to adapt to different corpora and question answering domains (such as bioinformatics).</Paragraph>
    <Paragraph position="2"> In addition to operating as a stand-alone answer typing component, our system can be combined with other existing answer typing strategies, especially in situations in which a catch-all answer type is used. Our experimental results show that ourprobabilisticmodeloutperformstheoracleand asystemusingautomaticnamedentityrecognition under such circumstances. The performance of our model is better than that of the semi-automatic system,whichisabetterindicationoftheexpected performance of a comparable real-world answer typing system.</Paragraph>
  </Section>
class="xml-element"></Paper>