<?xml version="1.0" standalone="yes"?>
<Paper uid="P06-1133">
  <Title>Are These Documents Written from Different Perspectives? A Test of Different Perspectives Based On Statistical Distribution Divergence</Title>
  <Section position="8" start_page="1061" end_page="1062" type="concl">
    <SectionTitle>
6 Conclusions
</SectionTitle>
    <Paragraph position="0"> In this paper we develop a computational test of different perspectives based on statistical distribution divergence between the statistical models of document collections. We show that the pro- null posed test can successfully separate document collections of different perspectives from other types of document collection pairs. The distribution divergence falling in the middle range can not simply be attributed to personal writing or speaking styles. From the plot of multinomial parameter difference we offer insights into where the different patterns of distribution divergence come from. Although we validate the test of different perspectives by comparing the DP condition with DT, SP, and ST conditions, the comparisons are by no means exhaustive, and the distribution divergence of some document collection pairs may also fall in the middle range. We plan to investigate more types of document collections pairs, e.g., the document collections from different text genres (Kessler et al., 1997).</Paragraph>
  </Section>
class="xml-element"></Paper>