🚀 go-pugleaf

*****************************************************************
Call For Participation

IJCAI'93 Workshop on a Problem-Solving Perspective on Explanation
*****************************************************************

The IJCAI Workshop on a Problem-Solving Perspective  on  Explana-
tion  will  be  held  during  the  Thirteenth International Joint
Conference on Artificial Intelligence in Chambery,  France.   The
one  day  workshop  will  bring  together  active  researchers in
several areas of automated explanation.  The goal of the workshop
is  to  study  the  explanation of knowledge-based systems from a
"problem-solving" perspective.  Explanation will be analyzed  for
impact from (1) the problem that is solved, (2) the model used to
represent the target artifact, and (3) the method used to  direct
the problem solving.

Explanation has grown from a by-product  of  the  development  of
knowledge-based  systems  to an important element of a knowledge-
based system's requirements specification.  Researchers and prac-
titioners  alike have realized, often in retrospect, that the ex-
planatory capabilities of a knowledge-based system are  dependent
on  a  combination  of  the problem being solved (e.g., diagnosis
versus design), the model used to represent the  artifact  (e.g.,
functional  versus  structural),  and  the  method used to direct
problem  solving  (e.g.,  generate-and-test   versus   cover-and-
differentiate).  The  result is  an explanation matrix indexed by
problem, model, and method types.   Although the specific indices
and dimensions may vary, a sample explanation matrix is shown be-
low.

                                                   PROBLEM
     
                                            !   ! C !   !   !   !
                                            ! D ! o !   !   !   !
                                            ! i ! n !   ! P ! C !
                                            ! a ! f !   ! r ! o !
                                            ! g ! i !   ! e ! n !
                                            ! n ! g ! P ! d ! t !
                                            ! o ! u ! l ! i ! r !
                                            ! s ! r ! a ! c ! o !
                                            ! e ! e ! n ! t ! l !
                    - - - - - - - - - - - - ---------------------
                          First Principles /   /   /   /   /   /|
                  - - - - - - - - - - - - --------------------- |
       MODEL                  Functional /   /   /   /   /   /  |
                - - - - - - - - - - - - ---------------------   |
                            Structural /   /   /   /   /   /    |
              - - - - - - - - - - - - ---------------------     /
                   Propose-and-Revise |   |   |   |   |   |    /
              - - - - - - - - - - - - ---------------------   /
              Cover-and-Differentiate |   |   |   |   |   |  /
              - - - - - - - - - - - - --------------------- /
                    Generate-and-Test |   |   |   |   |   |/
              - - - - - - - - - - - - ---------------------
     
                       METHOD

Many explanation systems have been developed within this  matrix.
However,  there has been  little study of the  explanation matrix
itself.   This workshop will focus on  analysis and comparison of
individual  cells  within this matrix.  For example, analysis may
show that the problem type (e.g., diagnosis) determines a set  of
things  that  need  to  be explained,  while the method and model
used to solve the problem (e.g., a cover-and-differentiate use of
causal knowledge) make some things easy or hard to explain.  Such
an analysis will lead to conclusions such as "diagnosis  requires
explanations  for  x,  y,  and z; the cover-and-differentiate ap-
proach using causal knowledge supplies  comparative  explanations
for x and y, but not  z.".

The workshop will  (1) collectively  propose a  matrix that leads
to  a useful and interesting partitioning of the explanation work
in knowledge-based systems; (2) propose and  evaluate  hypotheses
concerning the impact of matrix cells on explanatory ability; (3)
serve as a collection of case studies addressing the construction
of  explanation  facilities  within an explicit "problem-solving"
perspective.

Researchers interested in participating in the IJCAI Workshop  on
a  Problem-Solving  Perspective  on Explanation should submit six
(6) copies of a paper describing their  research  no  later  than
March  1,  1993.   Papers  must be original and are limited to 10
pages.  The title page of the paper must explicitly  include  the
problem, model, and method used to place the work in the explana-
tion matrix.  The specific indices, for example diagnosis  versus
design,  may  be any indices deemed appropriate by the authors of
the paper.  A "readings in" style book is anticipated as a result
of this workshop.

Important deadlines are as follows:

        submissions due       : March 1, 1993         
        author notification   : April 15, 1993         
        final papers received : June 15, 1993
        preprints distributed : July 15, 1993

Please send your submissions or inquires to:

Michael R. Wick
Computer Science Department
University of Wisconsin - Eau Claire
Eau Claire, WI 54702-4004
wick@csunix.cs.uwec.edu
1-715-836-2804 (office)
1-715-836-2923 (fax)

The Workshop Committee:

Patrick  Brezillon,  University  of  Paris  VI         
B. Chandrasekaran, Ohio State University
Rose Dieng, INRIA-Sophia-Antipolis
Anne Keuneke, California State University - Chico
Michael Tanner, George Mason University
Michael Wick(chair), University of Wisconsin - Eau Claire