<?xml version="1.0" standalone="yes"?>
<Paper uid="T78-1029">
  <Title>INDIRECT RESPONSES TO LOADED QUESTIONS*</Title>
  <Section position="9" start_page="207" end_page="207" type="concl">
    <SectionTitle>
CONCLUSION
</SectionTitle>
    <Paragraph position="0"> The problem of producing apparently intelligent behavior from a NL system has traditionally been viewed in Artificial Intelligence as a problem of modelling human cognitive processes, or modelling knowledge about the real world. It has been demonstrated here that such approaches must include a pragmatic theory of the conventions and properties of the use of language, to function effectively. Domain driven inferences must be complemented by language driven inferences to appropriately process NL. Further, it has been argued that language driven inference mechanisms help to control the inference process, and can provide a more general and computationally attractive solutions to many problems previously thought to require domain driven inference.</Paragraph>
    <Paragraph position="1"> A descriptive theory of one type of cooperative indirect response to inappropriate questions has been presented, and extended to a prescriptive (computational) theory by restricting the domain of application to DB query systems.</Paragraph>
    <Paragraph position="2"> This theory has been implemented using language driven mechanisms in the design of CO-OP, a cooperative query system. The result is the generation of appropriate corrective indirect responses in a computationally efficient and domain transparent fashion.</Paragraph>
  </Section>
class="xml-element"></Paper>