<?xml version="1.0" standalone="yes"?>
<Paper uid="W02-1411">
  <Title>Acquisition of Lexical Paraphrases from Texts</Title>
  <Section position="8" start_page="5" end_page="5" type="concl">
    <SectionTitle>
6 Conclusions
</SectionTitle>
    <Paragraph position="0"> We propose a process to acquire paraphrasing pairs of content words from a non-parallel raw corpus. We utilize contextual similarity, obtained from the corpus, to compute paraphrasability between any two content words.</Paragraph>
    <Paragraph position="1"> Some of the word pairs that unexpectedly have high paraphrasability are filtered out by using external linguistic knowledge such as proper nouns and antonyms. Moreover, our proposed heuristic, obtained through observation, can increase acquisition accuracy. These processes in combination are able to obtain more than 1,700 paraphrase pairs with approximately 66% accuracy. null Our interest in this research is not to pursue higher accuracy in automatic processing but to obtain any kind of paraphrasing knowledge as fast as possible. From this point of view, the coverage of the acquisition process is a more serious problem for us than accuracy. Our preliminary experiment showed that a drastic drop in accuracy is observed even if we increase coverage gradually. We need to find another filtering criterion to avoid this problem.</Paragraph>
  </Section>
class="xml-element"></Paper>