<?xml version="1.0" standalone="yes"?>
<Paper uid="W03-0406">
  <Title>Unsupervised learning of word sense disambiguation rules by estimating an optimum iteration number in the EM algorithm</Title>
  <Section position="8" start_page="5" end_page="5" type="concl">
    <SectionTitle>
7 Conclusions
</SectionTitle>
    <Paragraph position="0"> In this paper, we improved the EM method proposed by Nigam et al. for text classification problems in order to apply it to WSD problems. To avoid some failures in the original EM method, we proposed two methods to estimate the optimum iteration number in the EM algorithm.</Paragraph>
    <Paragraph position="1"> In experiments, we tested with 50 noun WSD problems in the Japanese Dictionary Task in SENSEVAL2. Our two methods greatly improved the original EM method. Especially, the score of noun evaluation words was equivalent to the best public score of this task. Furthermore, our methods were also effective for verb WSD problems. In future, we will tackle three works: (1) To find other effective features for unsupervised learning of verb WSD, (2) To improve the estimation method of the optimum iteration number in the EM algorithm, and (3) To investigate the reason for the failure of the EM method.</Paragraph>
  </Section>
class="xml-element"></Paper>