<?xml version="1.0" standalone="yes"?>
<Paper uid="W05-0808">
  <Title>A hybrid approach to align sentences and words in English-Hindi parallel corpora</Title>
  <Section position="6" start_page="62" end_page="62" type="evalu">
    <SectionTitle>
4 Results
</SectionTitle>
    <Paragraph position="0"> We performed manual evaluation of our word alignment algorithm on a set of parallel data aligned at the sentence level. The parallel texts consist of 3954 English and 5361 Hindi words taken from the EMILLE Corpus. We calculate our results in terms of the number of aligned English word groups. The precision is calculated as the ratio of the number of correctly aligned English word groups to the total number of English word groups aligned by the system, and recall is calculated as the ratio of the number of correctly aligned English word groups to the total number of English word groups created by the system. We obtained 77% precision and 67.79% recall for many-to-many word alignment. Figure 4.1 shows an example of the word alignment results.</Paragraph>
  </Section>
class="xml-element"></Paper>