<?xml version="1.0" standalone="yes"?>
<Paper uid="N04-1033">
  <Title>Improvements in Phrase-Based Statistical Machine Translation</Title>
  <Section position="10" start_page="0" end_page="0" type="relat">
    <SectionTitle>
9 Related Work
</SectionTitle>
    <Paragraph position="0"> Recently, phrase-based translation approaches became more and more popular. Some examples are the alignment template system in (Och et al., 1999; Och and Ney, 2002) that we used for comparison. In (Zens et al., 2002), a simple phrase-based approach is described that served as starting point for the system in this work. (Marcu and Wong, 2002) presents a joint probability model for phrase-based translation. It does not use the word alignment for extracting the phrases, but directly generates a phrase alignment. In (Koehn et al., 2003), various aspects of phrase-based systems are compared, e.g. the phrase extraction method, the underlying word alignment model, or the maximum phrase length. (Tomas and Casacuberta, 2003) describes a linear interpolation of a phrase-based and an alignment template-based approach.</Paragraph>
  </Section>
class="xml-element"></Paper>