<?xml version="1.0" standalone="yes"?>
<Paper uid="W01-1629">
  <Title>Spoken Dialogue Control Based on a Turn-minimization Criterion Depending on the Speech Recognition Accuracy</Title>
  <Section position="4" start_page="0" end_page="0" type="metho">
    <SectionTitle>
3 Experiments
</SectionTitle>
    <Paragraph position="0">  Weevaluatedtheproposedmethodbysimulation. In the simulation,the system conversed with a simulated user program. Simulation witha simulateduser enables rapidprototyping and evaluation (Eckert et al., 1998). The  conversationwasnotdonebyexchangingspoken language, but by exchanging attribute-value pairs.</Paragraph>
    <Paragraph position="1"> Simulated User Program The simulated user program works in the following steps:  1. Select a request. The request never changes throughout the dialogue 2. Tell thesystem therequest ora subset of the request 3. Respond YesorNoifthesystemconfirms 4. Give corrections at random if confirmation contains errors 5. Respond to the demand from the system 6. Tell the system that there is no infor null mation if the system refers to attributes with which the user is not concerned Specification of Test Task We prepared a fictitious task for simulation. This task accepts six types of user demand. There are six attributes,and twoof them have meaning dependence like the family name and department. Thenumbersofpersons,familynames, and departments are 3000, 1000, 300 respectively. null  complete a dialogue Comparison with a Naive Method For comparison,wepreparedanaiveconfirmation dialogue control method, with the following specifications: 1. If the user request can be fixed uniquely and there are unbound attributes requiredforthatrequest,demandthoseat- null tributes one by one.</Paragraph>
  </Section>
  <Section position="5" start_page="0" end_page="0" type="metho">
    <SectionTitle>
2. Iftherearevaluesthatarenotconfirmed,
</SectionTitle>
    <Paragraph position="0"> confirm them one by one.</Paragraph>
    <Paragraph position="1"> 3. Iftheuserrequesttypecan'tbefixedyet, demand a value for an attribute in the orderofthenumberofuserrequest types that require that attribute.</Paragraph>
    <Paragraph position="2"> Experimental Results Figures 1 and 2  showtheaveragenumberofturnsanditsvariance out of 1000 diaglogue. We can see from these figures that our method can complete dialoguesinshorterturnsthanothermethods under various levels of recognition accuracy. In addition, the variance is small in almost every range, which illustrates the stability of our method.</Paragraph>
  </Section>
class="xml-element"></Paper>