COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	basic.rep[e85,jmc]	Report for NSF on Basic Research in AI
C00005 ENDMK
C⊗;
basic.rep[e85,jmc]	Report for NSF on Basic Research in AI

	During the period of this grant John McCarthy continued
his research on formalizing common sense and on the development
of the circumscription method of non-monotonic reasoning and the
application of circumscription to the formalization work.
The main results are in the attached (McCarthy 1985) which was
presented at the AAAI Workshop on Non-monotonic Reasoning held
in October 1984 and which, after further revision, has been
submitted to {\it Artificial Intelligence} for publication.

	Circumscription was first described in (McCarthy 1980)
and applied to some examples of the kinds of non-monotonic reasoning
used in artificial intelligence.  Circumscripion makes it more
feasible to tackle the problem of making a database of general
common sense knowledge.  Such a database could be used by arbitrary
programs provided the knowledge was of a sufficiently general
character and represented in a general way.  It has been a
major difficulty for AI that the formalizations were all usable
only in specialized contexts.  Circumscription seems to permit
greater generality, but it had to be generalized as was done
in (McCarthy 1985).

	The problem of what is common sense has attracted
increased attention in AI, because it has become clear that
present expert systems are limited by its lack.  (McCarthy 1983)
addresses the problem of where expert systems need common sense
knowledge and reasoning ability.  McCarthy's retiring Presidential
Address to the American Association for Artificial Intelligence
further characterizes common sense knowledge and reasoning.  This
address is will be published in {\it AI Magazine} when the
written form is completed.