<?xml version="1.0" standalone="yes"?>
<Paper uid="H01-1036">
  <Title>Information Extraction with Term Frequenciesa0</Title>
  <Section position="4" start_page="0" end_page="0" type="evalu">
    <SectionTitle>
4. RESULTS
</SectionTitle>
    <Paragraph position="0"> In a large corpus there is duplicate or supporting information for almost any given question. The term frequency formula utilizes this knowledge, through two simple premises: the more a term is repeated, the more conceivable it is the correct answer, and the less likely a term appears by chance, the more probable it is also correct.</Paragraph>
    <Paragraph position="1"> The duplication component's importance in formula (1) can be evaluated by modifying the value of a63 in the term frequency equation: null</Paragraph>
    <Paragraph position="3"> Figure 2 demonstrates the value that duplicate information in the passages has on the result by modifying a63 .</Paragraph>
    <Paragraph position="4"> The graph reveals that as the importance of duplicate terms increases, the performance of the system strengthens. By eliminating the repetition part of the equation (a63 a5a50a65 ) the system only achieves a mean reciprocal rank of 0.237. As expected and demonstrated in the graph, the value of this part of the formula reaches a maximum before decreasing the overall system's accuracy.</Paragraph>
  </Section>
class="xml-element"></Paper>