Multi-agent systems interaction through intelligent routing services

This article presents and discusses a generic architecture that supports interaction and collaborative work among multiple communities of intelligent agents, addressing the problem of information overload and selective attention; i.e. the problem of how an agent (or an agent's community or multiple communities of agents) decide on which subsets of potentially large set of data should be attended to. Agents can be full-scale expert systems, simple computational platforms that support dialogues between the users and the agents, or even programs that emulate physical devices like sensors. Agents are selected according to their expertise and domain's knowledge, and integrated into the communities environments whenever are beneficial.