<?xml version="1.0" standalone="yes"?>
<Paper uid="C82-1020">
  <Title>NATURAL LANGUAGE ACCESS TO STRUCTURED TEXT</Title>
  <Section position="2" start_page="0" end_page="0" type="intro">
    <SectionTitle>
INTRODUCTION
</SectionTitle>
    <Paragraph position="0"> At SRI we are developing a system for natural language access to textual material, The system is to provide access to a textbook or other document of some importance, by returning relevant passages in response to a user's natural language request. Currently we are using the Hepatitis Knowledge Base, a compendium of current knowledge about hepatitis compiled by the National Library of Medicine, although the techniques we are devising are in no way particular to this document \[cf. Walker, 1982\]. The project has two phases. In the first, we are developing text access procedures for translating a user's request into an underlying logical form and, in order to locate the appropriate passages, matching the logical form with a Text Structure which expresses the structure of the document as a whole and summarizes the content of individual passages in terms of canonical predicates (Walker and Hobbs, 1981\]. In the second, longer-term effort, we are developing procedures for automatically generating portions of the Text Structure directly from the text.</Paragraph>
  </Section>
class="xml-element"></Paper>