Cooperation in a multi-agent community

Abstract

This paper describes a framework for cooperation suitable for being exploited in a society of cooperative distributed agents. We briefly present a tool UPShell - University of Porto Shell, which may be used to generate specific Intelligent Systems, although its main purpose is to build up a community of coarse-grain, semi-autonomous agents out either of those generated Intelligent Systems or of already pre-existent Knowledge Based Systems. After selecting the agents and the machines where each agent is going to run, UPShell sets up an architecture enabling agents to communicate and cooperate. Inside each agent, a blackboard-based architecture supports a flexible control and natural integration of different modules responsible for internal activity (problem solving) as well as external activity (cooperation and communication).