基于关系积的属性约简算法

粗糙集的属性约简是一个NP难问题,目前尚无高效的算法.基于集合理论,提出了关系积概念和基于关系积的属性约简算法,把决策表的属性约简过程转化为关系积的运算,减小了对决策表的扫描次数,提高了属性约简的效率;算法采用自底向上和宽度优先的搜索策略,可确保找到最小属性约简集.结合实例,给出了算法的具体实现.

Attribute reduction algorithm based on attribute union

Attribute reduction of rough sets is an NP hard problem, but there is not a popular efficient algorithm presently. The attribute union concept based on the set theory and the attribute reduced algorithm based on this concept were presented. The algorithm translates the attribute reduction to find the attribute union, reducing the number of scanning the decision table and improving the reduced efficiency. The scanning strategy from bottom to top and with width priority can insure to find the minimal reduction. Also, an example was presented to describe the algorithm.