作　　者： ; ;

机构地区： 湛江师范学院数学与计算科学学院

出　　处： 《计算机应用与软件》 2008年第1期58-60,共3页

摘　　要： 在利用差别矩阵求解决策表的相对核方法中,针对HU方法的错误,人们提出了各种各样新的差别矩阵及求相对核的方法,但计算代价高。把决策属性与条件属性放在一起构造出一个新的差别矩阵,得到了差别矩阵的若干性质和定理。在此基础上提出了求决策表的正区域、相对核、相对约简和最小约简的新算法,分析了该算法的时间复杂性。理论分析和实例表明,与现有的属性约简算法相比,该算法的时间复杂性较低。 In order to correct the error of HU method based on discemable matrix for computing the core of a decision table, some new discernable matrixes and the computation of a relative core are proposed, but the cost of computation is too high. A new discernable matrix consisting of condition attribute and decision attribute is presented, and several properties and theorems of the discernable matrix are gained. Based on the above results, a new algorithm for calculating positive region, relative core, relative reduction and minimal reduction of decision table is proposed. The time complexity of the algorithm is analyzed. The theoretical analysis and an example show that the time complexity of this algorithm is lower than that of the other present attribute reduction algorithms,

关 键 词： 差别矩阵 最简差别集 差别函数 相对核 相对约简 正区域

领　　域： [自动化与计算机技术] [自动化与计算机技术]