A development framework for distributed artificial intelligence

The authors describe distributed artificial intelligence (DAI) applications in which multiple organizations of agents solve multiple domain problems. They then describe work in progress on a DAI system development environment, called SOCIAL, which consists of three primary language-based components. The Knowledge Object Language defines models of knowledge representation and reasoning. The metaCourier language supplies the underlying functionality for interprocess communication and control access across heterogeneous computing environments. The metaAgents language defines models for agent organization coordination, control, and resource management. Applications agents and agent organizations will be constructed by combining metaAgents and metaCourier building blocks with task-specific functionality such as diagnostic or planning reasoning. This architecture hides implementation details of communications, control, and integration in distributed processing environments, enabling application developers to concentrate on the design and functionality of the intelligent agents and agent networks themselves.<
>