一种不确定性规则挖掘的混合方法及其应用

An Integrated Method of Mining Uncertain Decision Rules and It's Ap plication

为了在含有连续条件属性的决策表中挖掘和应用决策规则,通过采用互信息熵作为模糊聚类的截距值选定准则,对信息系统对象集进行模糊聚类划分,并根据粗糙集的有关概念,确定基于截距的近似集和边界,从中挖掘出不确定性决策规则,进一步建立决策知识库。最后以该知识库为识别模型,结合模糊隶属度的计算和不确定性推理进行地层识别,取得了令人满意的效果。

In order to mine decision-rules from decision table in which conditional attributes are continual,information entropy is used as a criterion to determine the intercept for fuzzy clustering.After the partition of the objects in the information system by fuzzy clustering,and according to the concepts of Pawlak's Rough Set,Rough Set based-on fuzzy intercept is redefined.Then uncertain rules are mined from decision table by use of approximate set and boundary set based-on fuzzy intercept.At last,it is satisfactory to recognize stratum with oil logging data by using uncertain rules,fuzzy subjection calculation and uncertain reasoning.