<?xml version="1.0" standalone="yes"?>
<Paper uid="A00-1030">
  <Title>Aggressive Morphology for Robust Lexical Coverage</Title>
  <Section position="4" start_page="219" end_page="219" type="concl">
    <SectionTitle>
4 Conclusions
</SectionTitle>
    <Paragraph position="0"> We have described an approach to robust lexical coverage for unrestricted text applications that makes use of an aggressive set of morphological rules to supplement a core lexicon of approximately 39,000 words to give lexical coverage that exceeds that of a much larger lexicon. This morphological analyzer is integrated with an extensive lexicon, an ontology, and a syntactic analysis system, which it both consults and augments. It uses ordered preferential rules that attempt to choose a small number of correct analyses of a word and are designed to deal with various states of lack of knowledge. When applied to 72 unknown words from a random sample of 100 distinct word types from the Brown corpus, its syntactic category assignments received a grade of B or better (using a grading system explained herein) for 97% of the words, and it correctly identified 95% of the root words. This performance demonstrates that one can obtain robust lexical coverage for natural language processing applications in unrestricted domains, using a relatively small core lexicon and an aggressive collection of morphological rules.</Paragraph>
  </Section>
class="xml-element"></Paper>