<?xml version="1.0" standalone="yes"?>
<Paper uid="A00-2028">
  <Title>Learning to Predict Problematic Situations in a Spoken Dialogue System: Experiments with How May I Help You?</Title>
  <Section position="3" start_page="210" end_page="210" type="intro">
    <SectionTitle>
2 Experimental System and Data
</SectionTitle>
    <Paragraph position="0"> HMIHY is a spoken dialogue system based on the notion of call routing (Gorin et al., 1997; Chu-Carroll and Carpenter, 1999). In the HMIHY call routing system, services that the user can access are classified into 14 categories, plus a category called other for calls that cannot be automated and must be transferred to a human operator (Gorin et al., 1997). ~ Each category describes a different task, such as person-to-person dialing, or receiving credit for a misdialed number. The system determines which task the caller is requesting on the basis of its understanding of the cMler's response to the open-ended system greeting A T~ T, How May I Help You?. Once the task has been determined, the information needed for completing the caller's request is obtained using dialogue submodules that are specific for each task (Abella and Gorin, 1999).</Paragraph>
    <Paragraph position="1"> In addition to the TASKSUCCESS dialogues in which HMIHY successfully automates the customer's call, illustrated in Figure 1, and the calls that are transferred to a human operator, there are three other possible outcomes for a call, all of which are problematic. The first category, which we call HANGUP, results from a customer's decision to hang up on the system. A sample HANGUP dialogue is in Figure 2. A caller may hang up because s/he is frustrated with the system; our goal is to learn from the corpus which system behaviors led to the caller's frustration.</Paragraph>
    <Paragraph position="2"> The second problematic category (WIZARD), resuits from a human customer care agent's decision to take over the call from the system. Because HMIHY is experimental, each call was monitored by a human 2 One potential use of our results is to provide better rules for making the transfer decision.</Paragraph>
    <Paragraph position="3"> SI: AT&amp;T How may I help you? UI: I need credit please.</Paragraph>
    <Paragraph position="4">  $2: Okay. What is the reason for the credit? U2: Miss, uh, Miss, different area code than I needed. $3: Sorry. I need to know whether you need credit for a wrong number, bad connection or a call that was cut off.</Paragraph>
    <Paragraph position="5"> U3: It is a wrong number. (impatient humming) $4: What was the number that you dialed?</Paragraph>
  </Section>
class="xml-element"></Paper>