<?xml version="1.0" standalone="yes"?>
<Paper uid="P98-2128">
  <Title>Learning Constraint Grammar-style disambiguation rules using Inductive Logic Programming</Title>
  <Section position="8" start_page="778" end_page="778" type="concl">
    <SectionTitle>
6 Conclusions
</SectionTitle>
    <Paragraph position="0"> Using the Progol ILP system, some 7 000 tag eliminating rules were induced from the Stockholm-UmePS Corpus. A lexicon was built from the corpus, and after lexicon look-up, test data (including only known words) was disambiguated with the help of the induced rules. Of 42 925 known words, 41 926 (98%) retained the correct reading after disambiguation. Some ambiguities remained in output: on an average 1.13 readings per word. Considering the experimental status of the tagger, we find the results encouraging. null</Paragraph>
  </Section>
class="xml-element"></Paper>