Cooperative agents in DISCIM environment

In complex domains, the decisions have to be based on information from various sources. Distributed environment DISCIM, based on multi-agent ideas, has been developed to support the necessary information exchange and distributed processing. This paper describes DISCIM specific methods allowing to create new agents from programs developed as stand-alone, originally. This is the case of a neural network emulator, of a qualitative reasoning system and of a data-base system POSTGRES — the “agentification” of all of them is treated in detail. Attention is devoted to comparison of the DISCIM system with other existing systems.