<?xml version="1.0" standalone="yes"?>
<Paper uid="N06-1052">
  <Title>Language Model Information Retrieval with Document Expansion</Title>
  <Section position="5" start_page="413" end_page="413" type="concl">
    <SectionTitle>
4 Conclusions
</SectionTitle>
    <Paragraph position="0"> In this paper, we proposed a novel document expansion method to enrich the document sample through exploiting the local corpus structure. Unlike previous cluster-based models, we smooth each document using a probabilistic neighborhood centered around the document itself.</Paragraph>
    <Paragraph position="1"> Experiment results show that (1) The proposed document expansion method outperforms both the no expansion baselines and the cluster-based models. (2) Our model is relatively insensitive to the setting of parameter M as long as it is suf ciently large, while the parameter a should be set according to the document length; short documents need a smaller a to obtain more help from its neighborhood. (3) Document expansion can be combined with pseudo feedback to further improve performance. Since any retrieval model can be presumably applied on top of the expanded documents, we believe that the proposed technique can be potentially useful for any retrieval model.</Paragraph>
  </Section>
class="xml-element"></Paper>