属性约简的新算法

摘要

Attribute reduction is one of the key problems for the rough sets theory. After the basic theories of the Rough Sets are studied,a new algorithm based on random information for decision -making system is presented. The new algorithm composes the attributes using the information of degree of dependency. And, it calculates the belief and plausibility functions of subset of attribute sets with an optimization order. The results show that the new algorithm can reduce the complexity of computation, especially when there is lots of condition attributes. The example has verified the effectiveness and superiority of the algorithm.%属性约简是粗糙集理论中一个核心研究问题,在对粗糙集中属性约简相关理论研究的基础上,提出了一种新的基于随机决策信息系统的属性约简算法.新算法充分利用属性依赖度所提供的信息对属性进行排序,并以一定的优化顺序来计算属性子集的信任函数或似真函数.计算结果表明:改进后的新算法计算量大大减小,尤其是当条件属性较多时,计算量的减少更加明显,从而大大提高了计算效率.计算实例验证了该算法的有效性,具有很强的优越性.