| Resumo: | Often agent-based applications, namely those related with diagnosis, control or classification problems, have to deal with unpredictable situations. This is the case where the agents must be able to call for effective problem solving strategies based on previous experiences. Indeed as result of past experiences and interaction, agents develop social behaviour (promoting inter-agent cooperation, mutual help or information exchange). Building on these ideas and going through models of human behaviour the design and development of a distributed multi-agent system environment was accomplished. This paper presents an overall description of the system, with emphasis on the expert agent classes, their architecture and models of interaction and cooperation.
|
|---|