<?xml version="1.0" standalone="yes"?>
<Paper uid="P98-1040">
  <Title>Dialogue Management in Vector-Based Call Routing</Title>
  <Section position="7" start_page="261" end_page="261" type="concl">
    <SectionTitle>
6 Conclusions
</SectionTitle>
    <Paragraph position="0"> We described and evaluated a domain independent, automatically trained call router that takes one of three actions in response to a caller's request. It can route the call to a destination within the call center, attempt to 6Gorin et al.'s results are measured without the possibility of system queries. To provide a fair comparison, we evaluated our muting module on all 389 calls in our test set using the scoring method described in (Gorin et al., to appear) (which corresponds roughly to our upperbound measure), and achieved a 94. ! % correct routing rate to 23 destinations when all calls are automatically routed (no false rejection), a substantial improvement over their system.</Paragraph>
    <Paragraph position="1"> formulate a disambiguating query, or route the call to a human operator. The routing module of the call router selects a set of candidate destinations based on n-gram terms extracted from the caller request and a vector-based comparison between these n-gram terms and each possible destination. If disambiguation is necessary, a yes-no question or wh-question is dynamically generated from among known n-gram terms in the domain based on closeness, relevance, and disambiguating power, thus tailoring the disambiguating query to the original request and the candidate destinations. Finally, our system performs substantially better than the best previously existing system, achieving an overall 93.8% correct routing rate for automatically routed calls when rejecting 10.2% of all calls.</Paragraph>
  </Section>
class="xml-element"></Paper>