<?xml version="1.0" standalone="yes"?>
<Paper uid="W06-3115">
  <Title>NTT System Description for the WMT2006 Shared Task</Title>
  <Section position="6" start_page="124" end_page="124" type="concl">
    <SectionTitle>
5 Conclusion
</SectionTitle>
    <Paragraph position="0"> We presented two translation models, a phrase-based model and a hierarchical phrase-based model. The former performed as well as the last year's best system, whereas the latter performed comparable to our phrase-based model. We are going to experiment new feature functions to restrict the too flexible reordering represented by our hierarchical phrase-based model.</Paragraph>
    <Paragraph position="1"> We also investigated different word alignment annotations, first using lower-cased corpus, second performing stemming, and third retaining only 4letter prefix. Differently preprocessed corpora resulted in quite divergent word alignment. Large phrase/rule translation tables were accumulated from three distinctly aligned corpora, which in turn, increased the translation quality.</Paragraph>
  </Section>
class="xml-element"></Paper>