<?xml version="1.0" standalone="yes"?>
<Paper uid="N04-4013">
  <Title>Web Search Intent Induction via Automatic Query Reformulation</Title>
  <Section position="8" start_page="0" end_page="0" type="concl">
    <SectionTitle>
7 Conclusion
</SectionTitle>
    <Paragraph position="0"> We have presented a method for providing useful suggested queries for underspeci ed informational queries.</Paragraph>
    <Paragraph position="1"> We evaluated our system using an unbiased metric against a standard web search system and found that our system enables users to more quickly nd relevant pages. This conclusion is based on an oracle assumption, which we also evaluate. Based on these evaluations, we can show that even under a pessimistic view point, our system outperforms the vanilla search engine.</Paragraph>
    <Paragraph position="2"> There is still room for improvement, especially in the predictivity results. We would like users to be able to more readily identify the class into which a relevant document (if one exists) would be found. We are investigating multi-document summarization techniques which might allow users to better pinpoint the category in which a relevant document might be found.</Paragraph>
  </Section>
class="xml-element"></Paper>