<?xml version="1.0" standalone="yes"?>
<Paper uid="I05-5009">
  <Title>Evaluating Contextual Dependency of Paraphrases using a Latent Variable Model</Title>
  <Section position="6" start_page="71" end_page="71" type="concl">
    <SectionTitle>
6 Conclusion
</SectionTitle>
    <Paragraph position="0">  Weproposedanevaluationmethodforthecontextual dependency of paraphrasing pairs using two latent variable models, pLSI and LDA. To evaluate a paraphrasing pair, we used sentences surrounding the given sentence as contextual information and approximated context by topics that correspond to a latent variable of a text model.</Paragraph>
    <Paragraph position="1"> The experimental results with paraphrases automatically extracted from a corpus showed that the proposed method achieved almost 60% accuracy.</Paragraph>
    <Paragraph position="2"> In addition, there is no major performance difference between pLSI and LDA. However, they have slightly different characteristics: LDA is robust against noisy sentences with long context, while pLSI is robust against information shortage due to short context. The results also revealed that any method's upper bound of accuracy using only contextual information is almost 77%.</Paragraph>
  </Section>
class="xml-element"></Paper>