<?xml version="1.0" standalone="yes"?>
<Paper uid="H01-1011">
  <Title>Automatic Title Generation for Spoken Broadcast News</Title>
  <Section position="7" start_page="0" end_page="0" type="concl">
    <SectionTitle>
4. CONCLUSION
</SectionTitle>
    <Paragraph position="0"> From the analysis discussed in previous section, we draw the following conclusions: 1. The KNN approach works well for title generation especially when overlap in content between training dataset and test collection is large.</Paragraph>
    <Paragraph position="1"> 2. The fact that NBL out performances NBF and TF.IDF out performance NBL and suggests that we need to distinguish important document words from those trivial words.</Paragraph>
  </Section>
class="xml-element"></Paper>