Fuzzy control of heterogeneous thinking of the hybrid intelligent multiagent system's agents

The paper is devoted to the development of the fuzzy inference subsystem for controlling the processes of collective heterogeneous thinking of agents of a hybrid intelligent multiagent system by an agent-facilitator. It allows the agent-facilitator to organize the work of the agents of the system in *accordance with S. Kaner's diamond of participatory decision-making model. The model reduces the group dynamics of the system into three sequential phases: the divergent thinking stage at which alternative solutions to the problem are developed; the groan stage at which the agent-facilitator uses methods that enhance “mutual understanding” between the agents; and the convergent thinking stage when the proposed alternatives are classified, ranked, and refined to make an agreed decision. Thanks to this mechanism, the relevance of the system to the practice of collective problem solving by experts under the guidance of a decision maker, with the mediation of a facilitator, increases. In combination with a hybrid component of an intelligent system and a multiagent approach underlying its architecture, modeling of collective heterogeneous thinking provides it with the ability to solve practical problems without significant simplification in a dynamic environment that limits the time to develop and make a decision.