摘要
在基于事例的推理中,通常采用判断相似度来进行事例检索。目前广泛采用的最近邻法存在着盲目判断、计算量大的缺陷,提出了一种改进的算法,采用聚类的方法把事例库分为合理的聚类,并找到每个聚类的均值,然后在推理中,新事例直接与每个均值进行比较,找到与它最相近的聚类,并在这个聚类中搜索最相近的事例。从而避免了盲目搜索,优化了算法。
Case-based reasoning can deal with those cases that are difficult to describe in exact rules, so it is very important in infer engine, and similarity is usually applied to search the case in it. At present, the nearest neighbor method is broadly adopted, but it has some shortcomings such as aimless searching, much work to calculate and so on. Therefore, a modified method is produced to reasonably decide the case using cluster algorithm, and to find out the mean for each duster. Then, the new case is compared directly with these means to find out the closest duster and the nearest case. Therefore aimless searching is avoided and the algorithm is optimized.
出处
《系统仿真学报》
EI
CAS
CSCD
北大核心
2005年第5期1045-1047,共3页
Journal of System Simulation
基金
航天科技集团"十五"预研课题(417010604-01)
关键词
聚类算法
最近邻法
基于事例推理
推理机制
cluster algorithm
nearest neighbor method
case-based reasoning
inference engine
