<?xml version="1.0" standalone="yes"?>
<Paper uid="W06-0503">
  <Title>Max-Planck-Institute for Computer Science</Title>
  <Section position="3" start_page="0" end_page="0" type="intro">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0"/>
    <Section position="1" start_page="0" end_page="0" type="sub_section">
      <SectionTitle>
1.1 Motivation
</SectionTitle>
      <Paragraph position="0"> Search engines, question answering systems and classification systems alike can greatly profit from formalized world knowledge. Unfortunately, manually compiled collections of world knowledge (such as e.g. WordNet (Fellbaum, 1998)) often suffer from low coverage, high assembling costs and fast aging. In contrast, the World Wide Web provides an enormous source of knowledge, assembled by millions of people, updated constantly and available for free. Since the Web data consists mostly of natural language documents, a first step toward exploiting this data would be to extract instances of given target relations. For example, one might be interested in extracting all pairs of a person and her birthdate (the birthdaterelation), pairs of a company and the city of its headquarters (the headquarters-relation) or pairs of an entity and the concept it belongs to (the instanceOf-relation). The task is, given a set of Web documents and given a target relation, extracting pairs of entities that are in the target relation. In this paper, we propose a novel method for this task, which works on natural language Web documents and does not require human interaction. Different from previous approaches, our approach involves a deep linguistic analysis, which helps it to achieve a superior performance.</Paragraph>
    </Section>
  </Section>
class="xml-element"></Paper>