<?xml version="1.0" standalone="yes"?>
<Paper uid="W06-3307">
  <Title>Integrating Co-occurrence Statistics with Information Extraction for Robust Retrieval of Protein Interactions from Medline</Title>
  <Section position="14" start_page="54" end_page="55" type="concl">
    <SectionTitle>
10 Conclusion
</SectionTitle>
    <Paragraph position="0"> Extracting relations from a collection of documents can be approached in two fundamentally different ways. In one approach, an IE system extracts relation instances from corpus sentences, and then aggregates the local extractions into corpus-level results. In the second approach, statistical tests based on co-occurrence counts are used for deciding if a given pair of entities are mentioned together more often than chance would predict. We have described  a method to integrate the two approaches, and given experimental results that confirmed our intuition that an integrated model would have a better performance. null</Paragraph>
  </Section>
class="xml-element"></Paper>