<?xml version="1.0" standalone="yes"?>
<Paper uid="W06-3116">
  <Title>Mood at work: Ramses versus Pharaoh</Title>
  <Section position="7" start_page="127" end_page="128" type="concl">
    <SectionTitle>
6 Discussion
</SectionTitle>
    <Paragraph position="0"> We presented a pairwise comparison of two decoders, RAMSES and PHARAOH. Although RAMSES is roughly twice as slow as PHARAOH, both de- null Test set [0,15] [16,25] [26,[?][</Paragraph>
    <Paragraph position="2"> Preferred indicates the preference of the subject (Pharaoh, Ramses or No preference). The column Improvedshows when a subject did prefer a translation and also said that the preferred translation was correct while the other one was not.</Paragraph>
    <Paragraph position="3"> coders offer comparable performances, according to automatic and informal human evaluations.</Paragraph>
    <Paragraph position="4"> Moreover, RAMSES is the product of clean framework: MOOD, a solid tool for research projects. Its code is open source and the architecture is modular, making it easier for researchers to experiment with SMT. We hope that the availability of the source code and the clean design of MOOD will make it a useful platform to implement new decoders.</Paragraph>
  </Section>
class="xml-element"></Paper>