<?xml version="1.0" standalone="yes"?>
<Paper uid="C04-1070">
  <Title>Using Bag-of-Concepts to Improve the Performance of Support Vector Machines in Text Categorization</Title>
  <Section position="9" start_page="0" end_page="0" type="concl">
    <SectionTitle>
8 Conclusions
</SectionTitle>
    <Paragraph position="0"> We have introduced a new method for producing concept-based (BoC) text representations, and we have compared the performance of an SVM classifier on the Reuters-21578 collection using both traditional word-based (BoW), and concept-based representations. The results show that BoC representations outperform BoW when only counting the ten largest categories, and that a combination of BoW and BoC representations improve the performance of the SVM over all categories.</Paragraph>
    <Paragraph position="1"> We conclude that concept-based representations constitute a viable supplement to word-based ones, and that there are categories in the Reuters-21578 collection that benefit from using concept-based representations.</Paragraph>
  </Section>
class="xml-element"></Paper>