A reference architecture for distributed intelligent systems and a preliminary description language for their integration

Artificial Intelligence and information systems research is just beginning to grapple with the problems of integrating autonomous, intelligent machine and human agents into complex problem-solving systems. These powerful multi-agent collaborative systems are theoretically possible, but conceptual tools for their development are lacking. This research develops a general model of distributed intelligent systems and a layered language for agent interaction in an attempt to provide a foundation for the realization of such conceptual tools. The model provides a basic definition of distributed, asynchronous, intelligent problem-solving systems. The language, which is constructed from temporal and modal logics, provides an axiomatized logical system in which the knowledge, beliefs, and intentions of multiple agents are exchanged. Combined, the language and model provide a reference architecture for the development of, and future research into distributed intelligent systems.