DEX:
An Expert System Shell for Multi-Attribute Decision Making

Purpose

DEX is an interactive computer program for the development of qualitative multi-attribute decision models and the evaluation of options. It is aimed at the support of decision-makers in complex decision-making problems: given a set of options (alternatives), the task is to select the option that best satisfies the goals of the decision maker.

DEX facilitates the following:

acquisition of decision models, which are composed of:
- a hierarchy of attributes, and
- the corresponding utility functions that are defined by decision rules;
consistency checking of decision rules;
acquisition, evaluation and analysis of (possibly incompletely defined) options;
explanation of the evaluation results;
group decision-making support.

Applied Methods

DEX is based on multi-attribute decision making. In this approach, the decision problem is decomposed into smaller, less complex subproblems. The decomposition is represented by a hierarchy (i.e., directed acyclic graph or, most commonly, a tree) of attributes. Options are evaluated by an aggregation that is gradually performed from the leaves towards the root of the hierarchy.

DEX differs from other multi-attribute decision support systems in that it uses qualitative (symbolic) attributes instead of quantitative (numeric) ones. Also, aggregation (utility) functions in DEX are defined by if-then decision rules rather than numerically by, for instance, the weighted sum. To evaluate incompletely or inaccurately defined options, DEX employs fuzzy or probabilistic distibutions of values. For the explanation of option evaluation, DEX primarily uses the method known as "selective explanation".

Availability

DEX runs under MS-DOS operating system on IBM PC and compatible computers. The program is implemented in Borland Pascal 7.0 and has approximately 15,000 lines of code.

The so-called "student version" of DEX is freely available for non-commercial applications. This version is fully functional, but limited to 25 attributes and 10 options. To install it, download the ZIP archive and unpack it to a directory of your choice. Also, see the READ_ME.TXT file contained in the archive.

Status

DEX was developed at the Department of Intelligent Systems, Jožef Stefan Institute, Ljubljana, Slovenia. The development started in 1987 and was, apart from very minor later revisions, completed in 1990. Although somewhat outdated, the program is still being used for practical applications. So far, it has been successfully applied in more than 50 realistic decision problems in various fields, such as:

selection and evaluation of computer hardware and software,
performance evaluation of enterprises and business partners,
project evaluation,
personnel management.

Documentation

M. Bohanec: DEX: An Expert System Shell for Multi-Attribute Decision Making, User's Manual (Software version 1.00ITR), Jožef Stefan Institute, Report DP-5896, 1990.

M. Bohanec: Introduction to DEX. Jožef Stefan Institute, Report DP-6240, 1991.

For students: a template for preparing a DEX Project Report (Microsoft Word document, RTF format, in Slovene).

Selected Publications

J.Efstathiou, V. Rajkovič (1979): Multiattribute Decisionmaking Using a Fuzzy Heuristic Approach. IEEE Transaction on Systems, Man and Cybernetics, Vol. SMC-9, No.6, June 1979.

M.Bohanec, V. Rajkovič (1990): DEX: An Expert System Shell for Decision Support. Sistemica 1(1), 145-157, 1990.

M. Bohanec, B. Urh, B., V. Rajkovič (1992): Evaluating options by combined qualitative and quantitative methods. Acta Psychologica 80, 67-89, North-Holland.

M. Bohanec, V. Rajkovič, B. Semolič, A. Pogačnik (1995): Knowledge-based portfolio analysis for project evaluation. Information & Management 28, 293-302, Elsevier.

M. Bohanec, V. Rajkovič (1995): Večparametrski odločitveni modeli. Organizacija 28(7), 427-438.

M. Bohanec, B. Cestnik, V. Rajkovič (1996): A management decision support system for allocating housing loans. In: Implementing systems for supporting management decisions (eds. P. Humphreys, L. Bannon, A. McCosh, P. Migliarese, J.-C. Pomerol), London: Chapman & Hall, 34-43.

M. Bohanec, B. Zupan, V. Rajkovič (1997): Hierarhični odločitveni modeli in njihova uporaba v zdravstvu. Zbornik CADAM-97: Računalniška analiza medicinskih podatkov (eds. I. Kononenko, T. Urbančič), Institut Jožef Stefan, 1-17.

Further Information

For further information about DEX and its applications contact its primary developer, Marko Bohanec. Any feedback about your experience with DEX will be greatly appreciated.