5. Multi-agent organization

Deploying ADMs on a large scale and in an open environment (such as the Web) requires structuring the interaction space and regulating the agents. In such a context, when numerous, possibly heterogeneous agents can enter and leave the system, problems of interoperability, efficiency and security arise: how can we guarantee mutual understanding between agents, facilitate their cooperation or protect the system against attacks by malicious agents? Critical application domains also require that the overall behavior of the system respects general constraints or properties (stability, compliance with commitments, pursuit of a strategy) or takes into account the organizational characteristics of the domain (hierarchy, procedure, regulation)