<?xml version="1.0" standalone="yes"?>
<Paper uid="P01-1039">
  <Title>Information Extraction From Voicemail</Title>
  <Section position="8" start_page="0" end_page="0" type="concl">
    <SectionTitle>
7 Conclusion
</SectionTitle>
    <Paragraph position="0"> In this paper, we have developed several techniques for extracting key pieces of information from voicemail messages. In contrast to traditional named entity tasks, we are interested in identifying just a selected subset of the named entities that occur. We implemented and tested three methods on manual transcriptions and transcriptions generated by a speech recognition system. For a baseline, we used a flex program with a set of hand-specified information extraction rules.</Paragraph>
    <Paragraph position="1"> Two statistical systems are compared to the baseline, one based on maximum entropy modeling, and the other on transducer induction. Both the baseline and the maximum entropy model performed well on manually transcribed messages, while the structure induction still needs improvement. Although performance degrades significantly in the presence of speech racognition errors, it is still possible to reliably determine the sound segments corresponding to phone numbers. null</Paragraph>
  </Section>
class="xml-element"></Paper>