<?xml version="1.0" standalone="yes"?>
<Paper uid="C04-1027">
  <Title>Learning theories from text</Title>
  <Section position="8" start_page="0" end_page="0" type="relat">
    <SectionTitle>
8 Related Work
</SectionTitle>
    <Paragraph position="0"> Rules such as the above express causality and interdependence between semantic predicates, which can be used to infer information for various linguistic applications. The idea of deriving inference rules from text has been pursued in (Lin and Pantel, 2001) as well, but that approach differs significantly from the current one in that it is aimed mainly at discovering paraphrases. In their approach text is parsed into paths, where each path corresponds to predicate argument relations and rules are derived by computing similarity between paths. A rule in this case constitutes an association between similar paths.</Paragraph>
    <Paragraph position="1"> This is quite different to the work currently presented, which provides more long range causality relations between different predicates, which may not even occur in adjacent sentences in the original texts. Other approaches such as (Collin et al., 2002) also aim to learn paraphrases for improving a Question-Answering system. Our work is perhaps more closely related to the production of causal networks as in (Subramani and Cooper, 1999), where the goal is to learn interdependency relations of medical conditions and diseases. In their work the dependencies only involve key words, but we believe that our techniques could be applied to similar biomedical domains to discover causal theories with richer inferential structure.</Paragraph>
  </Section>
class="xml-element"></Paper>