基于覆盖算法与多层前馈网络的案例库维护

Case-base maintenance based on covering algorithm and MFNN

针对运行在电子商务等交互式应用领域中的案例推理系统时,其案例库的规模很容易达到成千上万条且不能削减的特点,提出同时从两方面着手完成案例库维护:一方面用覆盖算法将案例库划分成覆盖领域,实现案例的选择性过滤使用;另一方面应用多层前馈神经网络改进案例匹配,提高检索效率.实验表明,该方法可以用来处理海量的高维数据,保证了系统的可用性.

Case-based reasoning （CBR） systems running in interactive domains like E-commerce, easily accumulate thousands of cases that are not reducible and thus presenting a problem to the efficiency of retrieval. To solve this problem, a new method was proposed to achieve case-base maintenance（CBM）from both sides： one was employing the alternative-covering algorithm to partition the case library to many covering domains, thus realizing selective filtering; the other was using multi-layer feedforward neural networks （MFNN） to deal with case retrieval within the large-scale case library. Experimental results indicate that the proposed method is especially feasible for processing vast and high dimensional data, which can effectively guarantee the system＇s usability and enhance its capability.