<?xml version="1.0" standalone="yes"?>
<Paper uid="W06-3107">
  <Title>Searching for alignments in SMT. A novel approach based on an Estimation of Distribution Algorithm [?]</Title>
  <Section position="8" start_page="51" end_page="53" type="concl">
    <SectionTitle>
6 Conclusions and Future Work
</SectionTitle>
    <Paragraph position="0"> In this paper, a new approach, based on the use of an Estimation of Distribution Algorithm has been presented. The results obtained with this technique are very promising even with the simple scheme here considered.</Paragraph>
    <Paragraph position="1"> According to the results presented in the previous section, the non-deterministic nature of the algorithm has not a real influence in the performance of this approach. Therefore, the main theoretical drawback of evolutionary algorithms have been proven not to be an important issue for the task we have addressed here.</Paragraph>
    <Paragraph position="2"> Finally, we are now focusing on the influence of these improved alignments in the statistical models for machine translation and on the degree of accu- null the alignments provided by the EDA. Search errors and model errors for EDA and GIZA++ algorithms are presented. In addition, the AER for the unidirectional EDA and reference alignments is also shown. These result are obtained on the Romanian-English 05 task  racy that could be achieved by means of these aligments. In addition to this, the integration of the aligment algorithm into the training process of the statistical translation models is currently being performed. null</Paragraph>
  </Section>
class="xml-element"></Paper>