Leonard N. Foner
Software agents have become increasingly popular for a variety of applications, many of which benefit from distribution on a network. However, many common approaches scale poorly in such an environment, because they assume the feasibility of knowing about all or most of any given agent’s peers. We investigate here some approaches and applications which will serve as a testbed for evaluating solutions to the above problems. These approaches involve clustering agents into clumps or "niches" that need only local knowledge; this work hence has relevance in the fields of learning, knowledge sharing, and distributed AI. Parts of the ideas above have been implemented, but many of them are speculative at this point, and feedback is actively encouraged.