<?xml version="1.0" standalone="yes"?>
<Paper uid="M93-1022">
  <Title>DEVELOPMENT EFFORT S NUBA was implemented in C++ . The software consists of three parts :</Title>
  <Section position="17" start_page="273" end_page="273" type="concl">
    <SectionTitle>
CONCLUSIONS
</SectionTitle>
    <Paragraph position="0"> Our long term goal is to develop a theory and a system for abduction-based natural language understanding .</Paragraph>
    <Paragraph position="1"> Our participation in MUC-5 took an important step towards this goal by demonstrating the power of a abduction-based semantic interpreter . Our official system is still incomplete : the parser is not integrated and the discourse module is only a crude first approximation . Nevertheless, NUBA's performance in formal testing is quite impressive . Our approach provides an interesting and perhaps better alternative to th e rule-based or pattern-based semantic interpretation .</Paragraph>
    <Paragraph position="2"> We are certainly not the first to realize the relevance and importance of abductive reasoning in natural language understanding . However, computational complexity plagued many previous abductive undrstanding systems. We have shown that with proper restriction on knowledge and explanation structure, abductiv e inference can be made very efficiently.</Paragraph>
  </Section>
class="xml-element"></Paper>