BISC-SIG on Decision Analysis

Christer Carlsson (ccarlsso@ra.abo.fi)
Wed, 19 Mar 1997 14:34:52 +0100


A NEW SPECIAL INTEREST GROUP IS BEING FORMED. OPEN FOR ALL BISC-ERS TO JO=
IN!
READ THIS ANNOUNCEMENT AND THEN LET US KNOW WHICH TASK FORCE YOU WANT TO
JOIN. PLEASE GIVE US YOUR IDEAS AND OFFER YOUR CONTRIBUTION TO THE STATE =
OF
ART PUBLICATIONS WE ARE GOING TO PRODUCE. PLEASE OFFER US SUPPORT OR GIVE=
US
BETTER IDEAS!

BISC-SIG ON DECISION ANALYSIS

1. Background.

The Special Interest Group on Decision Analysis will focus on several
different aspects of decision analysis:
. single criterion decision making under certainty, randomness an=
d
uncertainty;
. linear programming modelling (including variants for
transportation, scheduling, networks, goal programming, etc.) and
optimal decisions;
. multiple objective linear programming (including variants for
utility function trade off, reference points, interactive
bargaining among nondominated points, interdependence, etc.), multiple
criteria decision making (mcdm) multiple criteria decision aid
(mcda) and optimal decisions;
. decision support systems (DSS) and intelligent systems for
decision support as means for building optimal decision processes=
;
In all these approaches attempts have been made at fuzzyfying the standar=
d
modelling techniques, which is easy to do in a technical sense, but which
does not make much sense otherwise, as the modelling substance is missing
and it is not clear - from the perspective of a decision maker - why we n=
eed
the fuzzy set theory.

The SIG on DA will work on a new theory base for decision analysis to anc=
hor
it in a real life decision making context and work out a conceptual
modelling framework for better decision making with the help of soft
computing technology.

2. Problems to be tackled, goals to be attained.

The theory base is partly to be found in the literature. The base will
probably evolve in several stages: (i) state of the art surveys, (ii)
identifying sound key contributions to serve as cornerstones, (iii) picki=
ng
out obvious shortcomings and missing parts of the theory base, (iv)
proposing and validating new elements and theory segments, and (v) writin=
g a
proposed new theory base.

The elements and problems we will have to tackle include:
. some consistent conceptual framework(s) for tackling real life
decision problems using fuzzy modelling and approximate reasoning=
;
. modify, unify and expand fuzzy DA theory;
. develop a conceptual basis for fuzzy lp, molp, mcdm and mcda;
. work out a conceptual framework for using fuzzy logic in decisi=
on
support technology;

When the theoretical issues have been solved or resolved, the next step w=
ill
be to work out specifications for soft computing software which should be
designed to be instrumental for implementing the new DA approach in real
world decision making.

In order to get the methods and the software launched sets of good test
cases for each one of the areas listed above should be worked out and
offered as a way to both demonstrate the new methods and to offer easy wa=
ys
of questioning their substance and testing the new algorithms. Cases coul=
d
also be used to encourage people to find alternative methods and modellin=
g
techniques, which would both initiate debate and create a bandwagon effec=
t.
Finally, cases would help spread the new DA approach by offering good
teaching material for undergraduate programs.

Cases are not enough. Fairly early on the challenge of making sense to
corporate decision makers should be undertaken. This could be done by
tackling either classical $-saving problems in new, more effective ways o=
r
by working on problems for which classical optimisation techniques do not
make sense, but for which soft computing techniques would produce more
effective and more productive results than plain experience and intuition.
The problem is, of course, that the problems need to be worthwhile in $
terms and that these problems are not readily accessible for research peo=
ple
wanting to test and prove a new theory and an improved technology.

The problems we want to tackle and the goals we want to attain thus inclu=
de:
1. to find a soft computing paradigm for DA;
2. to create a consistent theory for "soft decision analysis";
3. to develop a methodology for optimal decision making with the
help of soft computing;
4. to create a recommended implementation program for soft DA and=
to
develop a taxonomy for successful implementation to be used for
follow up studies of implementation programs;

3. Action program.

The SIG on DA will be formed as a network of task forces with a minimum o=
f
3-5 members per task force. A task force will be formed for each of the
following areas:
. traditional single criterion decision making;
. linear programming modelling and optimal decisions;
. multiple objective linear, multiple criteria decision making
(mcdm) multiple criteria decision aid (mcda) and optimal decision=
s;
. decision support systems (DSS) and intelligent systems for
decision support;

Each task force will,
1. do an initial state-of-the-art report of its assigned area;
2. collect 3-5 obvious shortcomings or areas of potential
improvement with standard DA approaches;
3. work out a proposal for how these improvement could/should be
carried out;
4. work out solutions in detail with theory, models, solutions ca=
ses
and follow up taxonomy;

The task forces will report their results on a Web-site, at BISC workshop=
s,
through invited sessions at international meetings or through IFORS SPC:s.

The SIG on DA will collect reports in edited books on Kluwer (or
corresponding) and will work out implementation programs for the results =
in
co-operation with ERUDIT and kindred groups. When preliminary results are=
at
hand funding for large scale research programs will be applied for throug=
h
the ESPRIT V or corresponding programs.

4. Steering group.

The SIG on DA has the following steering group:

Christer Carlsson, IAMSR / Abo Akademi University (chairman)
Robert Full=E9r, Eotvos Lorand University
Hans Hellendoorn, Siemens Group
Masatoshi Sakawa, Hiroshima University
Lotfi Zadeh, BISC / UC Berkeley
Milan Zeleny, Fordham University

Web-site is maintained by Abo Akademi University with links to BISC, ERUD=
IT,
EURO WGs and the MCDA IG.

March, 1997

Christer Carlsson