<?xml version="1.0" standalone="yes"?>
<Paper uid="W04-0816">
  <Title>Anselmo Pe~nas Dpto. Lenguajes y Sistemas Inform'aticos UNED, Spain anselmo@lsi.uned.es</Title>
  <Section position="6" start_page="0" end_page="0" type="evalu">
    <SectionTitle>
5 Results at Senseval-3
</SectionTitle>
    <Paragraph position="0"> The results submited to Senseval-3 were generated with the system described in Section 4.4.</Paragraph>
    <Paragraph position="1"> Since one sense is assigned to every test context, precison and recall have equal values. Table 4 shows official results for the Lexical Sample Task at Senseval-3 in the four languages we have participated in: Spanish, Catalan, English and Italian.</Paragraph>
    <Paragraph position="2">  recall Differences between languages are quite remarkable and show the system dependence on the training corpora and the sense inventory. In the English task, 16 test instances have a correct sense not present in the training corpus. Since we don't use the dictionary information our system was unable to deal with none of them. In the same way, 68 test instances have been tagged as &amp;quot;Unasignable&amp;quot; sense and again the system was unable to detect none of them.</Paragraph>
  </Section>
class="xml-element"></Paper>