可信度差别矩阵及其属性约简

针对决策表中因存在不一致数据而造成求核和属性约简的错误,引入对象可信度的定义,并将其添加到决策表中,以区分不一致和一致对象;然后,给出差别矩阵的定义,以及基于该差别矩阵的核属性和属性约简定义,同时证明了由该定义获得的核和属性约简与正区域的核和属性约简是等价的;最后,设计属性约简算法及其改进算法,并通过实例和实验验证了算法的有效性。

Discernibility Matrix Based on Credibility and Attribute Reduction Method

Core attribute and attribute reduction are two of important parts researched in rough set theory. There exist the errors of computing the core and attribute reduction which caused by the inconsistency of decision table. To deal with above problem, firstly, the definition of the object credibility is introduced, which is added into the decision table to tell inconsistent objects and consistent objects. And then, a definition of the discernibility matrix is provided, and the definitions of the core and attribute reduction based the discernibility matrix are put forward. It is proved that the core and attribute reduction acquired from the definitions are the core and attribute reduction based on positive region. Finally, the general algorithm and its improved algorithm for computing attribute reduction are designed, and both of example and experiment results show that the algorithms are effiective.