<?xml version="1.0" standalone="yes"?>
<Paper uid="H91-1061">
  <Title>Evaluating Text Categorization I</Title>
  <Section position="6" start_page="317" end_page="317" type="concl">
    <SectionTitle>
CONCLUSION
</SectionTitle>
    <Paragraph position="0"> Text categorization plays a variety of roles in text-based systems. Evaluation of categorization effectiveness is important, both for confidence in operational systems and for progress in research. Several good measures, based on a model of binary decision making, are available for evaluating the effectiveness of a categorization. I have discussed some of the issues to consider in using these measures, and stressed that the purpose for which categorization is being done needs to be considered. The use of both indirect and direct evaluation is preferable. I also discussed how some of the work done by text extraction systems can be viewed as categorization and evaluated in a similar fashion, though new measures are needed as well.</Paragraph>
  </Section>
class="xml-element"></Paper>