COMMENT ⊗   VALID 00003 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	\input buslet[1,ra]
C00008 00003	\noindent Abstract: This is a proposal for the third year of NSF
C00009 ENDMK
C⊗;
\input buslet[1,ra]
\parindent 0pt
\parskip 8pt

\centerline{Proposal}
\centerline{Basic Research in Artificial Intelligence}
\bigskip
\centerline{This is a proposal renewal of the NSF grant DCR 84-14393}
\bigskip
\bigskip

\centerline{Accomplishments}

\disleft 15pt:{$\bullet$}:
During the second year of the proposed three year period McCarthy continued
his work on his circumscription formalism and its application for formalizing
common sense knowledge.  His paper ``Applications of circumscription
to formalizing common sense knowledge'' appeared in {\it Artificial
Intelligence}.  The issues concerning unintended models of the
circumscription formulas describing the effects of actions were
studied further.  Lifschitz has two proposed solutions of the problem,
Kautz another and Shoham a third.  In McCarthy's opinion none of the
solutions is entirely satisfactory in that they depend too much
on giving a special role to time and//or causality in the formalisms,
whereas humans do similar reasoning in cases not involving time.

\disleft 15pt:{$\bullet$}:
McCarthy has begun a study of the formalization of context.  Relations
$holds(p,c)$, and $value(exp,c)=a$, standing for proposition $p$ holds
in context $c$ and the value of expression $exp$ in context $c$ is $a$
were introduced and studied.  A relation $c1 ≤ c2$ standing for context
$c1$ is a specialization of context $c2$ was introduced.  The specializations
can involve either abbreviation, e.g. allowing $on(cat,mat)$
for $on(belong(McCarthy,cat),Mat367)$ or assumption, e.g. $c1$ specializes
the general English language context to a situation in which
wearing seat belts is required by law.  Non-monotonic reasoning is
used to move propositions up and down the context hierarchy.  It
appears that formalization of context will be required for a common
sense database.  The problem of constructing a general common sense database
presents acute problems for the logical (or other) representation
of common sense information in a way that can be used by arbitrary
programs written after the information is entered in the database.
Present representations of common sense information are always
specialized for the use of particular programs.

\bigskip
\centerline{Proposed Work}

\disleft 15pt:{$\bullet$}:
During the period January 1, 1987 to December 31, 1987, McCarthy will continue
his work on formalizing common sense knowledge, especially using
circumscription and pursuing the concept of mental situation calculus.

\disleft 15pt:{$\bullet$}:
The work on formalization of context will be continued.

\disleft 15pt:{$\bullet$}:
Formalisms suitable for a general database of common sense knowledge
will be developed.

\bigskip
\centerline{Recent and forthcoming publications}

{\bf McCarthy, John (1986)}:
``Applications of Circumscription to Formalizing Common Sense Knowledge''
{\it Artificial Intelligence}, April 1986
%  circum.tex[f83,jmc]

{\bf McCarthy, John (1987)}:
``Review of {\it Artificial Intelligence --- The Very Idea} by John
Haugeland'' to appear in {\it SIAM Journal}.
% haugel[f86,jmc]

{\bf McCarthy, John (1987)}:
``Generality in Artificial Intelligence'', to appear in
a collection of Turing lectures published by ACM
% genera[w86,jmc]

{\bf McCarthy, John (1987)}:
``Logic in Artificial Intelligence'', to appear in
 {\it Daedalus}.
% logic.2[w87,jmc]

\end

\end
\noindent Abstract: This is a proposal for the third year of NSF
grant DCR 84-14393.  During this year Professor John McCarthy will
continue his basic research in artificial intelligent.  The emphases
in this year will be on continuing the development non-monotonic reasoning methods
and beginning work on formalizing contexts.