<?xml version="1.0" standalone="yes"?>
<Paper uid="T78-1030">
  <Title>ON REASONING BY DEFAULT</Title>
  <Section position="10" start_page="216" end_page="216" type="concl">
    <SectionTitle>
5. CONCLUSIONS
</SectionTitle>
    <Paragraph position="0"> Default reasoning may well be the rule, rather than the exception, in reasoning about the world since normally we must act in the presence of incomplete knowledge. Moreover, aside from mathematics and the physical sciences, most of what we know about the world has associated exceptions and caveats. Conventional logics, such as first order logic, lack the expressive power to adequately represent the knowledge required for reasoning by default. We gain this expressive power by introducing the default operator.</Paragraph>
    <Paragraph position="1"> In order to provide an adequate formal (as opposed to heuristic) foundation for default reasoning we need a well articulated theory of default logic. This requires, in part, a theory of the semantics of default logic, a suitable notion of theoremhood and deduction, and conditions under which the default inference rules are effective and the set of theorems unique. Since in any realistic domain, all of the default schemata of Section 2 will be in force (together, no doubt, with others we have not considered) we require a deeper understanding of how these different schemata interact. Finally, there is an intriguing relationship between certain defaults and the complexity of the underlying representation. Both the closed world and frame defaults implicitly represent whole classes of first order axioms. Is this an accidental phenomemon or is some general principal involved?</Paragraph>
  </Section>
class="xml-element"></Paper>