<?xml version="1.0" standalone="yes"?>
<Paper uid="C86-1058">
  <Title>Context Analysis System for Japanese Text</Title>
  <Section position="3" start_page="246" end_page="246" type="concl">
    <SectionTitle>
6: Conclusion
</SectionTitle>
    <Paragraph position="0"> In the above sections we have proposed a system for extracting contextual information from natural language texts using a contextual representation structure as a knowledge structure. Our structure has proven itself useful for expressing contents of Japanese newspaper articles. Though we propounded the method used in our system to understand natural language texts in every field, some of its specifications such as the treatment of titles are oriented toward special\[ features of newspaper articles.</Paragraph>
    <Paragraph position="1"> At present, the applications of this system are restricted to stories dealing with terrorism. For these limits to be extended, the number of the contextual structures must be increased and the concept thesaurus scale enlarged. We believe -that the natural language understanding system described in this paper is flexible enough to allow for such extension. Computer facilities must, of course, also be taken into account.</Paragraph>
    <Paragraph position="2"> As our system is still in the development stage, some parts are not yet complete. Our dictionary is still rather small. For these reasons, &amp;quot;the scope of this paper has been limited to processing ability for a restricted category of newspaper articles.</Paragraph>
  </Section>
class="xml-element"></Paper>