<?xml version="1.0" standalone="yes"?>
<Paper uid="W05-0827">
  <Title>Improving Phrase-Based Statistical Translation by modifying phrase extraction and including several features</Title>
  <Section position="4" start_page="0" end_page="149" type="intro">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0"> Statistical Machine Translation (SMT) is based on the assumption that every sentence e in the target language is a possible translation of a given sentence f in the source language. The main difference between two possible translations of a given sentence is a probability assigned to each, which has to be learned from a bilingual text corpus. Thus, the translation of a source sentence f can be formulated as the search of the target sentence e that maximizes the translation probability P(e|f),</Paragraph>
    <Paragraph position="2"> 0This work has been supported by the European Union under grant FP6-506738 (TC-STAR project).</Paragraph>
    <Paragraph position="3"> If we use Bayes rule to reformulate the translation probability, we obtain,</Paragraph>
    <Paragraph position="5"> This translation model is known as the source-channel approach [1] and it consists on a language model P(e) and a separate translation model P(f|e) [5].</Paragraph>
    <Paragraph position="6"> In the last few years, new systems tend to use sequences of words, commonly called phrases [8], aiming at introducing word context in the translation model. As alternative to the source-channel approach the decision rule can be modeled through a log-linear maximum entropy framework.</Paragraph>
    <Paragraph position="8"> The features functions, hm, are the system models (translation model, language model and others) and weigths, li, are typically optimized to maximize a scoring function. It is derived from the Maximum Entropy approach suggested by [13] [14] for a natural language understanding task. It has the advantatge that additional features functions can be easily integrated in the overall system.</Paragraph>
    <Paragraph position="9"> This paper addresses a modification of the phrase-extraction algorythm in [11]. It also combines several interesting features and it reports an important improvement from the baseline. It is organized as follows. Section 2 introduces the baseline; the following section explains the modification in the phrase extraction; section 4 shows the different features which have been taken into account; section 5 presents the evaluation framework; and  the final section shows some conclusions on the experiments in the paper and on the results in the shared task.</Paragraph>
  </Section>
class="xml-element"></Paper>