<?xml version="1.0" standalone="yes"?>
<Paper uid="H94-1073">
  <Title>Assessing the Retrieval Effectiveness of a Speech :Retrieval System by Simulating Recognition Errors</Title>
  <Section position="5" start_page="371" end_page="371" type="concl">
    <SectionTitle>
4. Conclusions
</SectionTitle>
    <Paragraph position="0"> We have shown that speech retrieval is feasible to some extent even when the recognition performance is poor. It should be noted that a retrieval effectiveness which is moderate because of recognition errors may well be in the range of the retrieval effectiveness of the commonly used boolean retrieval method.</Paragraph>
    <Paragraph position="1"> In the case of MEDLARS, for instance, the boolean retrieval method achieves a retrieval effectiveness which corresponds to 40 % detection rate and 140 false alarms per indexing feature per hour \[3\]. It seems that recognition errors and occurrences of query features in the documents have different distributions and standard retrieval methods are quite good in distinguishing these two distributions. Further investigations are needed to study the influence of the length of the queries and the length of the documents.</Paragraph>
  </Section>
class="xml-element"></Paper>