<?xml version="1.0" standalone="yes"?>
<Paper uid="E06-2009">
  <Title>An ISU Dialogue System Exhibiting Reinforcement Learning of Dialogue Policies: Generic Slot-filling in the TALK In-car System</Title>
  <Section position="7" start_page="121" end_page="121" type="concl">
    <SectionTitle>
6 Conclusions and Future Work
</SectionTitle>
    <Paragraph position="0"> This report has described work done in the TALK project in building a software prototype baseline &amp;quot;Information State Update&amp;quot; (ISU)-based dialogue system in the in-car domain, with the ability to use dialogue policies derivedfrom machinelearningand also to perform onlinelearningthroughinteraction. We described the scenarios, gave a component level description of the software, and a feature level description and exam3We choose &amp;quot;orig city&amp;quot; because it is the least important and is already filled at the start of many COMMUNICATOR dialogues.</Paragraph>
    <Paragraph position="1"> ple dialogue.</Paragraph>
    <Paragraph position="2"> Evaluation of this system (i.e. comparing the system with hand-coded vs. learnt dialogue policies) is ongoing. Initial evaluation of learnt dialogue policies (Lemon et al., 2005; Henderson et al., 2005) suggests that the learnt policy performs at least as well as a reasonable hand-codedsystem (the TALK policy learnt for COMMUNICATOR dialogue management outperforms all the individual hand-coded COMMUNICATOR systems). null The main achievements made in designing and constructing this baseline system have been: a0 Combining learnt dialogue policies with an ISU dialogue manager. This has been done for online learning, as well as for strategies learnt offline.</Paragraph>
    <Paragraph position="3"> a0 Mapping learnt policies between domains, i.e.</Paragraph>
    <Paragraph position="4"> mapping Information States and system actions between DARPA COMMUNICATOR and in-carinformation seeking tasks.</Paragraph>
    <Paragraph position="5"> a0 Fragmentaryclarification strategies: the combination of ATK word confidence scoring with ISU-based dialoguemanagementrules allows us to explore word-based clarification techniques.</Paragraph>
  </Section>
class="xml-element"></Paper>