<?xml version="1.0" standalone="yes"?>
<Paper uid="W03-1211">
  <Title>Question Answering on a Case Insensitive Corpus</Title>
  <Section position="9" start_page="4" end_page="4" type="concl">
    <SectionTitle>
6 Conclusion
</SectionTitle>
    <Paragraph position="0"> An effective approach to perform QA on case insensitive corpus is presented with very little degradation (2.8%). This approach uses a high performance case restoration module based on HMM as a preprocessor for the NLP/IE processing of the corpus. There is no need for any changes on the QA system and the underlying IE engine which were originally designed for handling normal, case sensitive corpora. It is observed that the limited QA degradation is due to the limited IE degradation.</Paragraph>
    <Paragraph position="1"> An observation from the research of handling case insensitive questions is that question processing degradation has more serious consequence affecting the QA performance. The current case restoration training corpus is drawn from the general news articles which rarely contain questions. As a future effort, we plan to focus on enhancing the case restoration performance by including as many mixed-case questions as possible into the training corpus for case restoration.</Paragraph>
  </Section>
class="xml-element"></Paper>