Intelligent multicriteria decision support: Overview and perspectives

https://doi.org/10.1016/S0377-2217(98)00213-6Get rights and content

Abstract

This paper presents a survey of the history and the recent status of the Multicriteria Decision Support Systems (MCDSSs). The last generation of MCDSSs is based on the synergistic operation of Artificial Intelligence and Multicriteria Analysis (MA) approaches. The contribution of these two different approaches into an integrated MCDSS is analysed through this paper, which concludes providing the recent software developments in this field.

Introduction

Multicriteria Analysis (MA) consists the background of the Multicriteria Decision Support Systems (MCDSSs) which includes models, methods and approaches that aim to aid the Decision Makers (DMs) to handle semistructured decision problems with multiple criteria, where the components are transitional and the required information insufficient. The last twenty five years, MA presents important improvements which arle reflected from the progress of the four theoretical trends (see Bana e Costa, 1990; Pardalos et al., 1995, for instance), that are analysed below and presented in a schematic way in Fig. 1, Fig. 2, Fig. 3, Fig. 4 respectively.

The first trend is the Value System approaches (American School; Fishburn, 1970, 1972, 1982; French, 1993; Keeney and Raiffa, 1976; Keeney, 1992; Von Winterfeldt and Edwards, 1993) aiming to the construction of a value system that aggregates the DM's preferences on the criteria based on strict assumptions (complete and transitive preference relation). The estimated value system by this approach provides a quantitative way that leads the DM's in his/her final decision.

The French School (Roy, 1976, 1985, 1989, 1990; Roy and Bouyssou, 1993; Vincke, 1992; Brans and Mareschal, 1989; Vanderpooten, 1989) using a non compensatory approach aims to the construction of a relation (Outranking Relation) that allow the incomparability among the decision actions. The Outranking Relation approach is not bounded into a mathematical model but providing further exploitation and processes deduce to support the DM to conclude to a “good” decision.

The Disaggregation Aggregation approach (Jacquet-Lagréze, 1984; Jacquet-Lagréze and Siskos, 1982; Siskos, 1980; Siskos and Yannacopoulos, 1985; Siskos et al., 1993), which consists the third theoretical trend, aim to analyze the DM behavior and cognitive style. Special iterative interactive procedures are used, where the components of the problem and DM's global judgement policy are analysed and following are aggregating into a value system. The target of this approach is to aid the DM to improve his/her knowledge on the problem's state and his/her way of preferring that entail a consistent decision to be achieved.

The fourth theoretical trend is the Multiobjective Optimisation approach (Zeleny, 1974, 1982; Evans and Steuer, 1973; Zionts and Wallenious, 1976, 1983; Siskos and Despotis, 1989; Korhonen, 1990; Jaszkiewicz and Slowinski, 1995; Jacquet Lagrèze et al., 1987; Wierzbicki, 1992), which consists an extension of the Mathematical Programming one, aiming to solve problems where there are no discrete alternative actions and the objectives are more than one. The solution is estimated through iterative procedures which lead to: (a) achieving the satisfaction levels of the DM on the criteria or; (b) constructing a utility model of the DM that is used for the selection of the solutions that are assessed from a utility maximization procedure; or (c) a combination of the above two described methods.

Section snippets

The history of MCDSS

Almost the MCDSS can be different from each other, they have similarities according to their structure. Most of them consist of three main components: (a) the Data Subsystem, (b) the Model Subsystem and (c) the Dialogue Subsystem. The Data Subsystem is used for the data management (store, update, restore and process). The Model Subsystem includes the software that implements the multicriteria approach in a structural form. The Dialogue Subsystem provides the tools for the communication

The intelligent MCDSSs

The last few years quite a considerable number of researchers oriented their activities on the field of the Artificial Intelligence (AI). AI includes methods and techniques that provide the intelligent behavior to the computers that is to say (Dicken and Newquist, 1985):

  • Learn and understand from the experience.

  • Conclude in situation where exist fuzziness and uncertainty.

  • Use logic in order to conclude.

  • Use knowledge and experience to manipulate the environment.

  • Think and reason.

  • Understand and infer

Approaches of integrating ES and MCDSS

The name Intelligent MCDSS concerns the MCDSSs that include a kind of Intelligence into their procedures that is to say they include characteristics of the intelligent behavior (as they were described in Section 3). This can be done with the synergistic operation of the Multicriteria methods and the techniques of AI (Expert Systems, Neural Networks). A considerable amount of research work has been done concerning the design and the functionality of these two different approaches. Turban (1993)

Implementation of intelligent MCDSS

The idea of the Intelligent MCDSS have been presented from the early years of 80s and a considerable number of researchers had worked on this approach. Even then, only recently integrated Intelligent MCDSS have been implemented that utilise the most important characteristics of AI technology. The most important works concerning the integration of these two approaches (AI and MA), which have been used in real work applications, are presented in the following paragraphs.

AI techniques were used by

Conclusions

The integration of the AI techniques on the MCDSS provides important improvements of the systems capabilities that can be categorised in two different directions. The first concerns the improvements of the efficient use of the MCDSSs. The MCDSSs include powerful and sometimes complicated processes. This is the reason that MCDSSs require by the DM or DA deep knowledge on the MA scientific fields. The quality and the friendliness of the dialogues between DM-DA and the system ensures the liability

References (87)

  • BouwmanM.J.

    Human diagnostic reasoning by computers: An illustration from financial analysis

    Management Science

    (1983)
  • Brans, J.P., Mareschal, B., 1989. The PROMETHEE methods for MCDM, the PROMCALC, GAIA and Bankadviser Software. Vrije...
  • Dicken H.K. Newquist H.P. 1985. AI Trends '85. Scottsdale, AZ:DM...
  • DoukidisG.I. et al.
    (1992)
  • Edwards, J.S., 1990. Expert systems in management and administration - are they really different from decision support...
  • Elam, J.J., Konsynski, B., 1987. Using AI techniques to enhance the capabilities of model management systems. Decision...
  • EvansJ.P. et al.

    A revised simplex method for linear multiple objective programs

    Mathematical Programming

    (1973)
  • Fishburn, P., 1972. Mathematics of decision theory....
  • FishburnP.C.

    Utility theory for decision making

    (1970)
  • FishburnP.C.

    The Foundation of Expected Utility

    (1982)
  • FoxM.S. et al.

    ISIS - a knowledge based system for factory scheduling

    Expert Systems

    (1984)
  • FrenchS.

    Decision Theory: An Introduction to the Mathematics of Rationality

    (1993)
  • Henderson, J.C., 1987. Finding synergy between decision support systems and expert systems. Decision Sciences,...
  • HongJ. et al.

    Data and model management in a generalized MCDM-DSS

    Decision Science

    (1991)
  • Jacquet-LagrèzeE.

    PREFCALC: évaluation et décision multicitère

    Revue de l'Utilisateur de l'IBM PC

    (1984)
  • Jacquet-LagrèzeE.

    Interactive assessment of preferences using holistic judgement: The PREFCALC system

  • JaszkiewiczA. et al.

    The Light Beam Search: Outranking based interactive procedure for multiple-objective mathematical programming

  • Jelassi, M.T., 1986. MCDM: From "Stand-Alone" methods to integrated and Intelligent DSS. In: Proceedings of the VII-th...
  • JelassiM.T. et al.

    A framework for building an expert system for MCDM methods selection

  • Kastner, J., Apte, C., Giesmer, J., Hong, S.J., Karaugh, M., Mays E., Tozawa, Y., 1986. A knowledge-based consultant...
  • KeeneyR.L.
    (1992)
  • KeeneyR. et al.
    (1976)
  • KingD.

    Modeling and reasoning: Integrating decision support with expert systems

  • KleinM. et al.

    Expert Systems: A decision support approach with applications in management and finance

  • KoC. et al.

    Multiple Criteria Decision making and Expert Systems

  • KorhonenP.

    A multiple objective linear programming decision support system

    Decision Support Systems

    (1990)
  • LehnerP.E. et al.
  • LévineP. et al.

    PRIAM, an interactive method for solving multiple criteria problems

    European Journal of Operational Research

    (1986)
  • LévineP. et al.

    Knowledge representation by schemata in finance expert systems

    Theory and Decision

    (1989)
  • McLeod, K., Reeves, G.R., 1991. An application of the AXIS solution framework to multiple objective aggregate...
  • MichaelsenR.H.

    An expert system for tax planning

    Expert Systems

    (1987)
  • MurphyF. et al.

    An intelligent support for formulating linear programming

    Decision Support Systems

    (1986)
  • MatsatsinisN.F. et al.

    MARKEX: An intelligent decision support system for product development decisions

    European Journal of Operational Research

    (1999)
  • Cited by (76)

    • Riding with the Surging Tide: A Review of MCDM's Evolution

      2022, International Journal of Information Technology and Decision Making
    View all citing articles on Scopus
    View full text