<?xml version="1.0" standalone="yes"?>
<Paper uid="W06-2911">
  <Title>Applying Alternating Structure Optimization to Word Sense Disambiguation</Title>
  <Section position="16" start_page="83" end_page="83" type="ackno">
    <SectionTitle>
5 Conclusion
</SectionTitle>
    <Paragraph position="0"> With the goal of achieving higher WSD performance by exploiting all the currently available resources, our focus was the new application of the ASO algorithm in the multi-task learning configuration, which improves performance by learning a number of WSD problems simultaneously instead of training for each individual problem independently.</Paragraph>
    <Paragraph position="1"> A key finding is that using ASO with appropriate feature / problem partitioning, labeled examples of seemingly unrelated words can be effectively exploited. Combining ASO multi-task learning with ASO semi-supervised learning results in further improvements. The fact that performance improvements were obtained consistently across several languages / sense inventories demonstrates that our approach has broad applicability and hence practical significance.</Paragraph>
  </Section>
class="xml-element"></Paper>