基于拓扑邻域的序信息系统属性权重确定方法

A Method of Determining Weights Based on Topology Neighbourhood for Ordered Information System

在多属性决策中,属性权重是影响决策结果的重要因素,其可由决策群体根据偏好信息直接给出、也可根据数据分布差异性来确定。然而,属性之间的关联性往往被忽视,准确地刻画多个属性之间的相互关系,有利于获得较为客观的决策结果。针对序信息系统属性权重问题,本文基于AFS拓扑邻域构建属性模糊测度度量方法及相应的Choquet积分,并将其应用于学生成绩排名问题中。实例分析表明所构建方法能有效地利用多粒度信息来减少属性确定中的主观程度和体现属性间的关联度。

Attribute weight plays an important role in influencing the decision results in multiple attribute decision-making. Attribute weight can be given by decision group based on preference information and determined by the differences of data distribution. However, the correlation of attributes is usually ignored. In order to obtain the better decision results, the interrelation of multiple attributes should be reflected accurately. In this paper, a measurement method is established based on AFS topology neighbourhood, which is combined with Choquet integral to form a rank method. What's more, we apply it to the ranking of students' mark. The example illustrates that the proposed method can make a good use of multi-granular neighbourhood information to reduce the subjective degree of determining attribute and reflect the degree of association among attributes.