<?xml version="1.0" standalone="yes"?>
<Paper uid="W04-3253">
  <Title>Sentiment analysis using support vector machines with diverse information sources</Title>
  <Section position="8" start_page="0" end_page="0" type="concl">
    <SectionTitle>
7 Conclusion
</SectionTitle>
    <Paragraph position="0"> The method introduced in this paper allows several methods of assigning semantic values to phrases and words within a text to be exploited in a more useful way than was previously possible, by incorporating them as features for SVM modeling, and for explicit topic information to be utilized, when available, by features incorporating such values.</Paragraph>
    <Paragraph position="1"> Combinations of SVMs using these features in conjunction with SVMs based on unigrams and lemmatized unigrams are shown to outperform models which do not use these information sources. The approach presented here is flexible and suggests promising avenues of further investigation.</Paragraph>
  </Section>
class="xml-element"></Paper>