MUCO: a cognitive system for problem solving based on self-organization: a “multi-agents” implementation with blackboard

The paper concerns the self organization aspect of expert knowledge for problem solving. To model the cognitive system of an expert, we present a generic modelling frame which is called a Specialized Agent. A Specialized Agent is a set of finalized knowledge with particular functions: some of them allow to plan the solving process according to a decomposition in sub problems, some control the sub problems solving processes, and others give a final result. So, a Specialized Agent is an entity able to cooperate with others, for a particular problem solving situation. Thus, all the expert knowledge can be modelled by a collection of Specialized Agents, where each of them has to cooperate with others to solve a global problem. This modelling approach is then illustrated through a medical application: the MUCO project