<?xml version="1.0" standalone="yes"?>
<Paper uid="W06-3406">
  <Title>Improving &amp;quot;Email Speech Acts&amp;quot; Analysis via N-gram Selection</Title>
  <Section position="8" start_page="40" end_page="40" type="concl">
    <SectionTitle>
7 Conclusions
</SectionTitle>
    <Paragraph position="0"> In this work we considered the problem of automatically detecting the intents behind email messages using a shallow semantic taxonomy called &amp;quot;email speech acts&amp;quot; (Cohen et al., 2004). We were interested in the task of classifying whether or not an email message contains acts such as &amp;quot;propose a meeting&amp;quot; or &amp;quot;deliver data&amp;quot;.</Paragraph>
    <Paragraph position="1"> By exploiting contextual information in emails such as n-gram sequences, we were able to noticeably improve the classification performance on this task. Compared to the original study (Cohen et al., 2004), this representation reduced the classification error rates by 26.4% on average. Improvements of more than 30% were observed for some acts (Propose and Commit).</Paragraph>
    <Paragraph position="2"> We also showed that the selection of the top n-gram features via Information Gain revealed an impressive agreement with the linguistic intuition behind the different email speech acts.</Paragraph>
  </Section>
class="xml-element"></Paper>