Development of a multi-agent system for the support of group decisions utilizing argumentation and multicriteria methods

Abstract

Group decision making is one of the most important and frequently encountered processes within companies and organizations, both in the public and private sectors (Turban 1988). The understanding, analysis and support of this process is difficult, due to the ill-structured, dynamic environment and the presence of multiple Decision Makers (DMs); each DM has his or her own perceptions and views on how the problem should be handled and which decision should be made (Jelassi et al. 1990). Thanks to the developments in multicriteria decision making methodologies and the increasing popularity of computerized MCDM methods, scientists and professionals have been provided with a set of tools whose usage can be advantageous in solving problems with multiple criteria. However, it is evident that the effectiveness of such procedures when used by multiple DMs remains to be proven. This necessitates the use of practical aggregation methods to extend the existing MCDM methodologies, as well as the computing methodologies, to support group decision problems (Iz and Krajewski 1992). The use of Group Decision Support Systems (GDSSs) is crucial when multiple persons are involved in the decision making process, since each DM has his or her own perceptions of the context and the decision problem at hand. In environments of this kind, the occurrence of conflicts among the members of the decision-making group is frequent. This conflict is referred to as interpersonal conflict (Bogetoft and Pruzan 1991). Factors that contribute to the occurrence of interpersonal conflicts include different values and objectives, different criteria and preference relations, lack of communication support among the members of the decision-making group etc. Noori (1995) recognizes that, from a practical point of view, conflicting objectives among the members of a group often exist due to interpersonal differences and goal incongruities. In coping with interpersonal conflicts, the aim is to achieve consensus among the DMs; in such problems, Multi-Criteria Decision Aid (MCDA) methods may be a useful tool. As argued by Bui and Jarke (1986), MCDM/MCDA methods provide an elegant framework for three important GDSS tasks: (a) representing multiple viewpoints of a problem, (b) aggregating the preferences of multiple DMs according to various group norms and (c) organizing the decision process. The framework offered by MCDM is simple but structured, while the simplicity of its outputs makes communicating, coordinating and aggregating individual analyses in the group decision making process easier. Jarke (1986) states that MCDM methods can serve as formal tools for preference surfacing and aggregation, as well as negotiation and mediation, in both cooperative and non-cooperative decision situations. Thus, the multiple criteria process of a GDSS is a key aspect of the system, as it provides a structured and integrated framework for the assessment of alternatives and criteria and for solution compromise. Multiple agent (multi-agent) systems have become a valuable tool in the field of group decision making and, recently, their application has been extended in the sector of group decision making using multicriteria decision analysis techniques. For the decision making process in a multi-agent context, numerous techniques and methodologies have been proposed and implemented over the years, providing various solutions and approaches to the problem of group decision making. The term “interaction” rather than “argumentation-based negotiation” has been chosen, because, although this methodology did actually begin as argumentation-based negotiation (Sycara 1989b, Parsons et al. 1998), it has now branched out and evolved into a completely unique type of multi-agent interaction in its own right, overcoming and surpassing the limitations of game-theoretic negotiation. It appears to be a very viable and promising methodology, because of the close and efficient approximation of the procedure used by human decision makers it achieves. This thesis proposes a Group Decision Support methodology and software system that attempts to support a group of decision makers bestowed with the solution of a choice problem (i.e. a problem of the problematic α), beginning from a set of individual ordinal rankings and using a combination of a heuristic algorithm and an argumentation protocol for the building of a consensus among the decision makers. The heuristic algorithm serves as a method of accelerating the argumentation-based negotiation on the alternatives. This thesis is part of the 03ED375 research project, implemented within the framework of the “Reinforcement Programme of Human Research Manpower” (PENED) and co-financed by National and Community Funds (75% from E.U.-European Social Fund and 25% from the Greek Ministry of Development-General Secretariat of Research and Technology).