Cohesive Hybrid Intelligent Multi-Agent System Architecture

Abstract

The paper is devoted to the development of architecture of a new class of intelligent system, namely a cohesive hybrid intelligent multi-agent system. This intelligent system is designed to model collective problem solving by an expert team at round-table. Like traditional hybrid intelligent multi-agent systems, this system integrates the advantages of the hybrid intelligent system concept and the multi-agent approach. The key difference of the proposed class of intelligent systems is the modeling of expert team cohesion, by coordinating goals, domain models and developing problem-solving protocol by agents. These mechanisms are especially relevant when a hybrid intelligent multi-agent system is built from agents developed by various independent teams. The paper presents the model of cohesive hybrid intelligent multi-agent system, its functional structure, and the architectures of its agents.