<?xml version="1.0" standalone="yes"?>
<Paper uid="W04-3228">
  <Title>Dependencies vs. Constituents for Tree-Based Alignment</Title>
  <Section position="7" start_page="0" end_page="0" type="concl">
    <SectionTitle>
6 Conclusion
</SectionTitle>
    <Paragraph position="0"> We present a comparison of constituent and dependency models for tree-to-tree alignment. Despite equalizing some mismatches in tree structure, the dependency representation does not perform as well, likely because it is less robust to large differences between the tree structures.</Paragraph>
    <Paragraph position="1"> Acknowledgments We are very grateful to Rebecca Hwa, Hao Zhang, everyone at the 2003 John Hopkins speech and language summer research workshop, and EMNLP's reviewers for their assistance, criticism, and data. This work was partially supported by NSF ITR IIS-09325646, NSF research infrastructure grant EIA-0080124, and NSF grant 0121285 to the summer workshop.</Paragraph>
  </Section>
class="xml-element"></Paper>