This article presents an agent-based fuzzy cognitive map (ABFCM) developed injecting the concept of multi-agent system (MAS) into the fuzzy cognitive map (FCM). Fuzzy cognitive map is used for qualitative modeling and simulation. Compared to the FCM, the ABFCM enables different inference algorithms in each node enabling simulation of systems with diverse behavior concepts. Each map node can exhibit individual, more or less intelligent, behavior and still can interact with other nodes to conclude on system behavior. Resulting method also enables automatic results interpretation adding additional intelligence to a classic FCM. Explanation of the obtained system architecture with FCM and MAS integration is presented in the article. The experimental results in the article are obtained with the ABFCM prototype, developed on the basis of ABFCM structure given in the article. Multi-agent technology can bring new properties into existing fields and methods, like in the ABFCM case.

Computer science