Un Nuevo Método de Explotar Relaciones de Preferencia Borrosas en Agentes de Decisión

The normative approach for decision-making is the dominant paradigm in designing intelligent decision agents. We discuss here the advantages of a more flexible way based on fuzzy logic. However, most exploitation methods of fuzzy preference relations do not provide good prescriptions. Recently some approaches based on the idea of reducing inconsistencies using evolutionary multiobjective optimization have been proposed. In this work a new method is presented based on similar ideas but improving them. The multiobjective optimization problem is separated into two steps and solved with a better model of preferences, also using a simpler evolutionary algorithm. These improvements allow us to achieve better solutions in a simpler way than the previous methods.