<?xml version="1.0" standalone="yes"?>
<Paper uid="C02-1100">
  <Title>Lenient Default Unification for Robust Processing within Unification Based Grammar Formalisms</Title>
  <Section position="8" start_page="77" end_page="77" type="concl">
    <SectionTitle>
7 Conclusion
</SectionTitle>
    <Paragraph position="0"> We proposed two new types of default unification, ideal and non-ideal lenient default unification. Ideal lenient default unification is desirable in that it maximizes the amount of information in the result, while other existing types of default unification maximize the amount of information in the default.</Paragraph>
    <Paragraph position="1"> Although non-ideal lenient default unification gives a less informative result than the ideal one, it works efficiently and retains the desiderata the ideal one satisfies.</Paragraph>
    <Paragraph position="2"> We also proposed a new approach to extend the coverage of a grammar. We extracted grammar rules from the results of robust parsing using lenient default unification. A series of experiments showed that the extracted rules work robustly, and the coverage of the XHPSG grammar for Penn Treebank greatly increased with a little overgeneration.</Paragraph>
  </Section>
class="xml-element"></Paper>