<?xml version="1.0" standalone="yes"?>
<Paper uid="C04-1060">
  <Title>Syntax-Based Alignment: Supervised or Unsupervised?</Title>
  <Section position="7" start_page="0" end_page="0" type="concl">
    <SectionTitle>
6 Conclusion
</SectionTitle>
    <Paragraph position="0"> We present a side-by-side comparison of syntactically supervised and unsupervised tree-based alignment, along with the non tree-based IBM Model 4.</Paragraph>
    <Paragraph position="1"> For Chinese-English, using trees helps the alignment task, but a data-derived tree structure gives better results than projecting automatic English parser output onto the Chinese string. The French-English task is easier overall, and exhibits smaller differences between the systems.</Paragraph>
    <Paragraph position="2"> Acknowledgments We are very grateful to Rebecca Hwa for assistance with the Chinese-English data, and to everyone who helped make the resources we used available to the research community. This work was partially supported by NSF ITR IIS-09325646.</Paragraph>
  </Section>
class="xml-element"></Paper>