<?xml version="1.0" standalone="yes"?>
<Paper uid="W04-0607">
  <Title>Feeding OWL: Extracting and Representing the Content of Pathology Reports</Title>
  <Section position="7" start_page="0" end_page="0" type="relat">
    <SectionTitle>
5 Related Work
</SectionTitle>
    <Paragraph position="0"> Acquisition of information from texts especially from the medical domain is a lively research area.</Paragraph>
    <Paragraph position="1"> Among the many projects in that field, we share some of our central concerns with the medSyndiKAte system (Hahn et al., 2002): robust text analysis of medical reports; a background knowledge base for guiding the analysis and storing the text's content; emphasis on handling co-reference phenomena. What distinguishes LUPUS from medSyndiKAte, though, is foremost the parsing scheme: the language used in the reports analysed by Hahn et al.</Paragraph>
    <Paragraph position="2"> is much closer to 'natural' language in that it contains sentences with tensed verbs. Accordingly, they use a variant of dependency parsing which is driven by verb information. As described in Section 2.2 above, this is not an option for us, given the style of our input texts, and hence our data renders a bottom-up chart parsing approach much more promising.</Paragraph>
    <Paragraph position="3"> Besides this difference, the work in medSynDiKAte predates the emergence of XML/web ontology standards and thus uses an earlier description logic knowledge representation language; we are hoping that by using a standard we will be able to allow even future semantic web technologies to work with our data.</Paragraph>
    <Paragraph position="4"> As for the robust analysis side, (Grover et al., 2002), also use a similar preprocessing pipeline in combination with parsing. However, they also focus on more &amp;quot;natural&amp;quot; input texts (Medline abstracts), and they use statistical rather than symbolic/ontology based methods for computing the meaning of compound nouns.</Paragraph>
  </Section>
class="xml-element"></Paper>