Orlando Belo, Departamento de Informatica, Escola de Engenharia, Universidade do Minho, Portugal
Jose Neves, Departamento de Informatica, Escola de Engenharia, Universidade do Minho, Portugal
Software agents can be grouped according to their application domains and expertise. When face to a problem, agents may cooperate exchanging resolution strategies or results, acting as unique and highly specialized entities. However, when isolated, are like "islands" of experts, with skills and knowledge reduced to their own areas of expertness, and confined to their environments. An agent may not obtain in its own world all the information it needs. But, acting according to its "culture" and the community's own rules of engagement, the agents may request such data from another world. Cooperation, exchange of information and knowledge privacy, involving different agents' worlds with distinct environments, are some of the matters discussed in this paper. A framework that allows the creation of computational tools, the development of new learning and coordination machinery, meeting the demands of the environment and the requirements of the individual agents, is presented.
Distributed Artificial Intelligence, Multi-Agent Systems, Blackboard Based Systems, Prolog