<?xml version="1.0" standalone="yes"?>
<Paper uid="W04-0838">
  <Title>SenseLearner: Minimally Supervised Word Sense Disambiguation for All Words in Open Text</Title>
  <Section position="6" start_page="0" end_page="0" type="evalu">
    <SectionTitle>
4 Evaluation
</SectionTitle>
    <Paragraph position="0"> The SENSELEARNER system was evaluated on the SENSEVAL-3 English all words data - a data set consisting of three texts from the Penn Tree-bank corpus, with a total of 2,081 annotated content words. Table 1 shows precision figures for each part-of-speech (nouns, verbs, adjectives), and contribution of each word class toward total recall.</Paragraph>
    <Paragraph position="1">  The average precision of 64.6% compares favorably with the &amp;quot;most frequent sense&amp;quot; baseline, which was computed at 60.9%. Not surprisingly, the verbs seem to be the most difficult word class, which is most likely explained by the large number of senses defined in WordNet for this part of speech.</Paragraph>
  </Section>
class="xml-element"></Paper>