<?xml version="1.0" standalone="yes"?>
<Paper uid="P04-3002">
  <Title>Improving Domain-Specific Word Alignment for Computer Assisted Translation</Title>
  <Section position="8" start_page="21" end_page="21" type="concl">
    <SectionTitle>
5 Conclusion
</SectionTitle>
    <Paragraph position="0"> This paper proposes an approach to improve domain-specific word alignment through alignment adaptation. Our contribution is that our approach improves domain-specific word alignment by adapting word alignment information from the general domain to the specific domain. Our approach achieves it by training two alignment models with a large-scale general bilingual corpus and a small-scale domain-specific corpus. Moreover, with the training data, two translation dictionaries are built to select or modify the word alignment links and further improve the alignment results.</Paragraph>
    <Paragraph position="1"> Experimental results indicate that our approach achieves a precision of 83.63% and a recall of 76.73% for word alignment on a user manual of a medical system, resulting in a relative error rate reduction of 21.96%. Furthermore, the alignment results are applied to a computer assisted translation system to improve translation efficiency.</Paragraph>
    <Paragraph position="2"> Our future work includes two aspects. First, we will seek other adaptation methods to further improve the domain-specific word alignment results.</Paragraph>
    <Paragraph position="3"> Second, we will use the alignment adaptation results in other applications.</Paragraph>
  </Section>
class="xml-element"></Paper>