<?xml version="1.0" standalone="yes"?>
<Paper uid="W05-0818">
  <Title>LIHLA: Shared task system description</Title>
  <Section position="8" start_page="113" end_page="113" type="concl">
    <SectionTitle>
4 Concluding remarks
</SectionTitle>
    <Paragraph position="0"> This paper has presented a lexical alignment method, LIHLA, which aligns words and multi-word units based on initial statistical word-to-word correspondences and language-independent heuristics. null In the experiments carried out at the &amp;quot;Shared task on word alignment&amp;quot; which took place at the Workshop on Building and Using Parallel Texts during ACL2005, LIHLA has been evaluated on English-Inuktitut and Romanian-English parallel texts achieving an AER of 22.72% and 44.49%, respectively.</Paragraph>
    <Paragraph position="1"> As future work, we aim at investigating the impact of using additional linguistic information (such as part-of-speech tags) on LIHLA's performance. Also, as a long-term goal, LIHLA will be part of a system implemented to learn transfer rules from sequences of aligned words.</Paragraph>
  </Section>
class="xml-element"></Paper>