<?xml version="1.0" standalone="yes"?>
<Paper uid="H01-1061">
  <Title>Robust Knowledge Discovery from Parallel Speech and Text Sources</Title>
  <Section position="5" start_page="0" end_page="0" type="concl">
    <SectionTitle>
3. SUMMARY
</SectionTitle>
    <Paragraph position="0"> Our goal is to align collections of stories from multiple text and speech sources in more than one language and then develop methods that exploit the resulting parallelism both as a tool to improve recognition accuracy and to enable the development of systems that can reliably extract information from parallel sources. Much like a teacher rephrases a concept in a variety of ways to help a class understand it, the multiple sources, we expect, will increase the potential of success in knowledge extraction. We envision techniques that will operate repeatedly on multilingual sources by incorporating newly discovered information in one language into the models used for all the other languages. Applications of these methods extend beyond news sources to other multiple-source domains such as office email and voice-mail, or classroom materials such as lectures, notes and texts.</Paragraph>
  </Section>
class="xml-element"></Paper>