A Mulriple Case-Based Reasoning Approach

Case-Based Reasoning (CBR), a branch of Knowledge-Based Expert System, is a rapidly growing reasoning method in the research of Artificial Intelligence(AI).It is applied in those fields where experiences can be provided rather than general rules.This paper presents a mulriple CBR model on the basis of typical CBR model.This model includes three case-bases contrasting with only one in the typical CBR system: Users case-base,Objects case-baes and application case-base,they work together to finish tasks.The aim of creating Users case-base for users of CBR system is to decide appropriate working strategy for different users when it can significant improve the speed and precision of search.Objects case-base is used to get arguments and characteristic information of current object.Application case-base working together with the information that come from user case-base and object case-base will give the problem solution.This paper also analyses some problems accompanying with this model and advises some simple and efficient solutions.As a result of making use of the theory about MCBR,we discuss the way using MCBR with mulriple case-based resoning system that improve the ability of auto-adaption when the local case-base is sparse.