扩展的粗糙集模型及其不确定性量度

为克服标准的粗糙集模型不能表示数据对象的不同重要性和属性的不同特性的局限 ,需对其进行扩展。在可变精度粗糙集的基础上 ,构造了一种新的扩展粗糙集模型。它通过在知识表示系统和决策表中引入数据对象的权值函数和属性的特性函数来克服上述局限。给出了适于数据对象具有不同重要性情况下的粗糙决策规则集合的不确定性量度 ,以其作为规则评价的标准 ,可以方便地融入主观偏好、先验知识等因素。通过对一个数据集的分析实例对此进行了说明

Generalized rough set model and its uncertainty measure

Some generalizations are needed to remedy the limitations of standard rough set models. A new generalized rough set model is constructed to deal with situations where different data objects have different importance and different attributes have different characteristics using a weighting function for data objects and the attribute characteristic function. This generalized model is used to extend an entropy based uncertainty measure by taking the weight function into account. The method can facilitate combining of factors such as the decision maker's preferences and prior domain knowledge in the rules evaluation processes. The experimental results on a real data set illustrate the advantages of the proposed method.