<?xml version="1.0" standalone="yes"?>
<Paper uid="H94-1039">
  <Title>RECENT IMPROVEMENTS IN THE CMU SPOKEN LANGUAGE UNDERSTANDING SYSTEM</Title>
  <Section position="3" start_page="0" end_page="0" type="intro">
    <SectionTitle>
1. INTRODUCTION
</SectionTitle>
    <Paragraph position="0"> Understanding spontaneous speech presents several problems that are not found either in recognizing read speech or in parsing written text. Since the users are not familiar with the lexicon and grammar used by the system, it is very difficult for a speech understanding system to achieve good coverage of the lexicon and grammar that subjects might use. Spontaneous speech often contains ungrammatical constructions, stutters, filled pauses, restarts, repeats, interjections, etc. This causes problems both for the recognizer and the parser.</Paragraph>
    <Paragraph position="1"> Stochastic language models tend to produce more robust recognition than grammar based models. They can be smoothed to allow for unseen word sequences and their scope is short enough to &amp;quot;get back on track&amp;quot; after an error. The parsing and understanding component also must be robust to the phenomena in spontaneous speech and to recognition errors. Even though the speech is disfluent and gramatically ill-formed, the relevant information is still consistent most of the time. We therefore try to model the information in an utterance rather than its grammatical structure. The natural language component of our system is oriented toward the extraction of information relevant to a task, and seeks to directly optimize the correctness of the extracted information (and therefore the system response). We use a flexible frame-based parser, which parses as much of the input as possible. This approach leads both to high accuracy and robustness.</Paragraph>
    <Paragraph position="2"> We have implemented a version of this system for the ARPA Air Travel Information Service (ATIS) task. Users are asked to perform a task that requires getting information from an Air Travel database.</Paragraph>
    <Paragraph position="3"> They must interact with the system by voice to find a solution. In this paper, we describe recent improvements in our system resulting from our efforts to increase the coverage given a limited amount of  In addition, we improved the basic performance of the parser and added a rejection mechanism.</Paragraph>
  </Section>
class="xml-element"></Paper>