【摘要】 最近相邻策略是基于事例推理(CBR)中常用的检索策略。针对该方法的缺点该文提出了变权值的最近相邻检索,并探讨了变权值带来的问题,在此基础上该文给出了两种解决方法,事例检索记忆表和采用神经网络与最近相邻策略相结合的方法来检索相似源事例,可在变权值的情况下快速地检索出相关的源事例。从而解决了事例库的设计者和使用者之间的由视角不同而产生的矛盾。更多还原

【Abstract】 The Nearest Neighbor is commonly used in Case-Based Reasoning system.In this paper the nearest neighbor of varied weights is proposed to meet the demand of retrieving cases in different point of views.The order of cases has been changed,the nearest neighbor strategy would have lower efficiency in retrieving the similar source cases in the case base,so some related problems of varied weights are discussed.Two methods of solving the varied-weight cases-retrieving problem are given which realize rapid case retrieval under the condition of varied weights,one is memory table of case retrieval,and the other is the nearest neighbor based on artificial neural network.So the inconsistency between the designers and the users are partly solved.The methods have the advantage of quick retrieving speed,especially for larger case base.更多还原