On-line multi-CBR agent dispatching

Abstract

It is well known that CBR is a fast and efficient problem-solving technique. However, keeping a single CBR system seems to be outdated and not practical. With the increasing requests for useful references for all kinds of problems and from different locations, multi-CBR agents located in different places are of great support to fast meet these requests. In this paper, first, a multi-CBR agents system is proposed. In this system, each CBR agent locates at different places, and is assumed to have the same ability to deal with new case (problem) independently. Next, when requests in a request queue from different places are coming one by one, we propose a new policy of dispatching which agent to orderly satisfy the request queue. In the whole paper, we assume that the system must solve one request by considering only past requests. In this context, the performance of a kind of greedy algorithm is not satisfied. We apply a new but simple approach-competitive algorithm for on-line problem (called ODAL) to decide the dispatching policy from the aspect of keeping comparative low cost. The corresponding on-line dispatching algorithm is proposed and the competitive ratio is given, based on which the dispatching of multi-CBR agents is optimized.