<?xml version="1.0" standalone="yes"?>
<Paper uid="W01-0713">
  <Title>Unsupervised Induction of Stochastic Context-Free Grammars using Distributional Clustering</Title>
  <Section position="10" start_page="0" end_page="0" type="concl">
    <SectionTitle>
9 Conclusion
</SectionTitle>
    <Paragraph position="0"> In conclusion, distributional clustering can form the basis of a grammar induction algorithm, by hypothesising sets of rules expanding the same non-terminal. The mutual information criterion proposed here can filter out spurious constituents.</Paragraph>
    <Paragraph position="1"> The particular algorithm presented here is rather crude, but serves to illustrate the effectiveness of the general technique. The algorithm is computationally expensive, and requires large amounts of memory to run efficiently. Though the results presented here are preliminary, I have shown how an unsupervised grammar induction algorithm can induce at least part of a linguistically plausible grammar from a large mixed corpus of natural language.</Paragraph>
  </Section>
class="xml-element"></Paper>