Similarity measure of agents’ ontologies in cohesive hybrid intelligent multi-agent system

Abstract

The work is devoted to the issues of building cohesive hybrid intelligent multi-agent systems that implement the following mechanisms for ensuring the joint work of heterogeneous agents: coordination of goals and agent ontologies, development of the problem solving protocol. These mechanisms are similar to the processes that occur in long-standing expert teams. In them, experts, interacting in the course of solving problems, exchange views, knowledge, take into account each other’s goals, establish formal and informal norms of communication. In the paper, in particular, a method for estimating the similarity of ontologies is considered, which is the part of the method for estimating the cohesion of agents of the hybrid intelligent multi-agent system. The estimation of cohesion is used by system’s agents as the interaction effectiveness criterion when making decisions about the need to converge goals, ontologies or correction of the interaction protocol.