<?xml version="1.0" standalone="yes"?>
<Paper uid="W05-0704">
  <Title>Examining the Effect of Improved Context Sensitive Morphology on Arabic Information Retrieval</Title>
  <Section position="7" start_page="31" end_page="31" type="concl">
    <SectionTitle>
5 Conclusion
</SectionTitle>
    <Paragraph position="0"> The paper investigated the effect of improved morphological analysis, especially context sensitive morphology, in Arabic IR applications compared to other statistical morphological analyzers and light stemming. The results show that improving morphology has a dramatic effect on IR effectiveness and that context sensitive morphology slightly improved Arabic IR over non-context sensitive morphology, increasing IR  The processing of the TREC collection using the in-context IBM-LM required 16 hours on a 2.4 GHz Pentium 4 machine with 1 Gigabyte of RAM compared to 10 minutes to perform light stemming.</Paragraph>
    <Paragraph position="1"> effectiveness by approximately 3%. The improvement is almost statistically significant. Developing better morphology could lead to greater retrieval effectiveness, but improving analyzers is likely to be difficult and would require careful determination of the proper level of conflation. In overcoming some of the difficulties associated with obtaining &amp;quot;better&amp;quot; morphology (or more fundamentally the proper level of word conflation), adaptive morphology done on a per query term basis or user feedback might prove valuable. Also, the scores that were used to rank the possible analyses in a statistical morphological analyzer may prove useful in further improving retrieval. Other IR techniques, such as improved blind relevance feedback or combination of evidence approaches, can also improve monolingual Arabic retrieval.</Paragraph>
    <Paragraph position="2"> Perhaps improved morphology is particularly beneficial for other IR applications such as cross-language IR, in which ascertaining proper translation of words is particularly important, and  in-document search term highlighting for display to a user.</Paragraph>
  </Section>
class="xml-element"></Paper>