信息表相对属性约简的一个算法

给出一种相对属性约简的算法 .该算法对于信息表中由决策属性决定的划分中的每个基本集合 ,计算它们的下逼近集并对每个条件属性在下逼近集中的取值进行检索 ,剔除可能是多余的条件属性 ,得到一个候选属性约简 .在此基础上 ,通过逐步扩展属性 ,最终得到一个属性约简 .与已有的一些算法相比 ,该算法计算量较少

An algorithm for relative reduction of attributes in information system

We present in this paper an algorithm for relative reduction of attributes in an information system. The algorithm first computes the lower approximation of every basic set in the equivalence class defined by the decision attribute, then, by retrieving the values taken by each condition attribute in the lower approximation sets, deletes some possibly superfluous attributes and finally obtains a relative reduction through extension of attributes. The algorithm requires less computational effort than some existing reduction algorithms.