<?xml version="1.0" standalone="yes"?>
<Paper uid="W06-3322">
  <Title>Extracting Protein-Protein interactions using simple contextual features</Title>
  <Section position="4" start_page="120" end_page="120" type="ackno">
    <SectionTitle>
4 Results
</SectionTitle>
    <Paragraph position="0"> Tables 1 and 2 show the results for the two models described above. The system achieves a peak per3These patterns are in regular expression form, i.e. &amp;quot;P1 word{0,n} Iverb word{0,m} P2&amp;quot;. This particular pattern matches sentences where a protein is followed by an iWord that is a verb, with a maximum of n words between them, and following this by m words maximum is another protein. In their paper, (Plake et al., 2005) optimise the values for n and m using Genetic Algorithms, but I will simply set them all to 5, which is what they report as being the best unoptimized setting.</Paragraph>
    <Paragraph position="1"> formance of 59.2% F1, which represents a noticeable improvement over previous results on the same dataset (52% F1 (Plake et al., 2005)), and demonstrates the feasibility of the approach adopted.</Paragraph>
    <Paragraph position="2"> It is seen that simple contextual features are quite informative for the task, but that a significant gains can be made using more elaborate methods.</Paragraph>
  </Section>
class="xml-element"></Paper>