Abstract: Discernibility matrix is an important method for knowledge reduction in rough set theory, the number of elements in discernibility matrix will directly influence the computation efficiency of knowledge reduction algorithm based on discernibility matrix method, in view of the time complexity and space complexity of these algorithms are not good, the main reason is that computing and storing the discernibility matrix of decision information system consumes a large of time and the discernibility matrix include many empty elements. For this issue, new construction method for discernibility matrix is proposed, which can delete all empty elements in discernibility matrix. On these bases, an algorithm for knowledge reduction in decision information system is presented, the algorithm can effectively consider core attribute in decision information system, which further improve the efficiency of the algorithm. Finally, an example result shows the feasibility of the proposed algorithm.

Key words: decision information system, core attribute, discernibility matrix, rough set

摘要： 差别矩阵方法作为求解粗糙集知识约简的关键技术之一，而差别矩阵中的元素个数将直接影响知识约简算法的计算效率，针对现有基于差别矩阵方法的知识约简算法的不足，并且当决策信息系统中样本量较大、决策类别数较少时，算法构造的差别矩阵中将存在大量空值元素。提出了一种新的差别矩阵构造方法，有效地剔除了差别矩阵中的空值元素，在此基础上，设计了一种决策信息系统的知识约简算法，由于算法能有效地利用核属性，进一步缩小了知识约简算法的效率，并通过算例分析说明了算法的可行性。

关键词: 策信息系统, 知识约简, 核属性, 粗糙集