<?xml version="1.0" standalone="yes"?>
<Paper uid="W04-0802">
  <Title>The SENSEVAL-3 Multilingual English-Hindi Lexical Sample Task</Title>
  <Section position="7" start_page="0" end_page="0" type="evalu">
    <SectionTitle>
6 Results
</SectionTitle>
    <Paragraph position="0"> All systems attempted all of the test instances, so precision and recall are identical, hence we report  the single Accuracy figure. Tables 2 and 3 show results for the t and ts subtasks, respectively. We note that the participating systems all exceeded the baseline (majority) classifier by some margin, suggesting that the sense distinctions made by the translations are clear and provide sufficient information for supervised methods to learn effective classifiers.</Paragraph>
    <Paragraph position="1"> Interestingly, the average results on the ts data are higher than the average results on the t data, which suggests that sense information is likely to be helpful for the task of targeted word translation. Additional investigations are however required to draw some final conclusions.</Paragraph>
  </Section>
class="xml-element"></Paper>