基于粒计算的信息系统规则提取方法

Rules extraction method of information systerm based on granular computing

根据关系的粒矩阵表示和矩阵运算简便直观的特点,计算条件属性集的粒关系包含度矩阵,并在该矩阵中获得满足给定阈值的决策规则集合;在保持该集合中元素个数不变的前提下对决策信息系统进行属性约简,并在约简的属性集上提取具有更高泛化能力的决策规则.理论证明和实例分析检验了该方法的可行性和有效性.

According to the simple and intuitive features of a relation represented by a matrix and matrix operations,an inclusion degree matrix of the granular-relations is calculated in the set of all condition attributes,and the set of the decision rules that meeting given threshold value is extracted in this matrix.Then,the attribute of decision-making information system is reduced without changing the numbers of elements in the set.Finally,the decision rules with higher generalization ability are extracted based on the reduced attribute set.The theoretical and example analysis verified the feasible and effectiveness of this method.