基于论域压缩的启发式属性约简算法

Heuristic Algorithm of Attribute Reduction Based on Universe Compression

论证在简化的信息系统上进行属性约简的可行性,指出某信息系统属性约简快速算法的计算结果可能含冗余属性,且在时间复杂度计算上存在错误。在此基础上,提出一种基于论域压缩的启发式属性约简算法,将相对支持度作为启发信息,缩小搜索空间,加入二次约简过程以消除冗余属性。实例分析表明,该算法具有较好的约简效果。

This paper gives a proof to the feasibility of the heuristic attribute reduction in simplified information system. It points out the results of one quick algorithm may contain redundancy attributes and time complexity calculation is wrong. This paper proposes a heuristic algorithm of attribute reduction based on universe compression. The relative support degree is used as heuristic information for reducing the searching room. A process of secondary reduction is used in order to eliminate redundancy attributes. A real example results demonstrate the improved algorithm has good reduction effect.