<?xml version="1.0" standalone="yes"?>
<Paper uid="W01-1602">
  <Title>VariantTransduction: A Method for Rapid Developmentof InteractiveSpoken Interfaces</Title>
  <Section position="2" start_page="0" end_page="0" type="intro">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0"> Developing non-trivial interactivespoken language applications currently requires signicant eort, often several person-months. A major part of this eort is aimed at coping with variation in the spoken language input by users. One approach to handling variation is to write a large natural language grammar manually and hope that its coverage is sucientformultiple applications (Dowding et al., 1994). Another approach is to create a simulation of the intended system (typically with a human in the loop) and then record users interacting with the simulation.</Paragraph>
    <Paragraph position="1"> The recordings are then transcribed and annotated with semantic information relating to the domain;; the transcriptions and annotations can then be used to create a statistical understanding model (Miller et al., 1998) or used as guidance for manual grammar development (Aust et al., 1995).</Paragraph>
    <Paragraph position="2"> Building mixed initiativespoken language systems currently usually involves the design of semantic representations specic to the application domain. These representations are used to pass data between the language processing components: understanding, dialog, conrmation generation, and response generation. However, such representations tend to be domain-specic, and this makes it dicult to port to new domains or to use machine learning techniques without extensive hand-labeling of data with the semantic representations. Furthermore, the use of intermediate semantic representations still requires a nal transduction step from the intermediate representation to the action format expected by the application back-end (e.g. SQL database query or procedure call).</Paragraph>
    <Paragraph position="3"> For situations when the eort and expertise available to build an application is small, the methods mentioned above are impractical, and highly directed dialog systems with little allowance for language variability are constructed.</Paragraph>
    <Paragraph position="4"> In this paper, we describe an approachto constructing interactivespoken language applications aimed at alleviating these problems. We rst outline the characteristics of the method (section 2) and what needs to be provided by the application builder (section 3). In section 4 and section 5 we explain variant expansion and the operation of the system at runtime, and in section 6 we describe how conrmation requests are produced by the system. In section 7 wegive some initial experimental results on varying the number of examples used to construct a call-routing application.</Paragraph>
  </Section>
class="xml-element"></Paper>