<?xml version="1.0" standalone="yes"?>
<Paper uid="J99-2002">
  <Title>at Asheville</Title>
  <Section position="9" start_page="204" end_page="204" type="concl">
    <SectionTitle>
9. Summary
</SectionTitle>
    <Paragraph position="0"> This paper has described an important class of probability models and procedures for model selection that have not been widely used in NLP. The procedures complement the set of available methods for balancing expressiveness and feasibility. The framework is understandable, powerful, and computationally feasible. Its effectiveness for an NLP problem is demonstrated here in a large-scale word sense disambiguation experiment.</Paragraph>
  </Section>
class="xml-element"></Paper>