混合语义下的不完备多尺度决策系统及其最优尺度选择

Incomplete Multi-Scale Decision System with Mixed Semantics and Its Optimal Scale Selection

现有不完备多尺度信息系统的定义存在局限性,即只考虑了遗漏型未知属性值并且认为对象在某个属性下缺失则在所有尺度下全部缺失.本研究从未知属性值的两种语义解释出发,定义混合语义下的不完备多尺度信息系统.为保持混合语义下的不完备多尺度信息系统尺度间的偏序关系,根据粒信息转换函数填补数据并扩展特征关系模型.研究在不同尺度下信息粒度的表示及其相互关系,并进一步定义基于扩展后特征关系的集合的上、下近似,讨论了它们的性质.在混合语义下的不完备多尺度决策系统中构建序贯三支决策模型,并基于该模型给出最小化不确定性的最优尺度选择方法.实验结果验证了数据填补和特征关系模型扩展的有效性,并且表明通过序贯三支决策模型能够选择不确定性最小的尺度作为最优尺度.

The definition of the existing incomplete multi-scale information system has limitations.Namely,only the missing unknown value is considered,and objects unknown at an attribute are missing at all scales of the attribute.First of all,the incomplete multi-scale information system with mixed semantics is defined in this paper based on the two kinds of semantic interpretations of unknown value.Secondly,in order to maintain the partial order relation between scales in incomplete multi-scale information systems with mixed semantics,the data is supplemented according to the granular information transformation function and the characteristic relation model is extended.Then,the representation of information granularity at different scales and their interrelations are given.The concepts of upper and lower approximation of sets based on extended characteristic relation are further defined,and their properties are discussed.Finally,a sequential three-way decisions model is constructed in an incomplete multi-scale decision system with mixed semantics,and an optimal scale selection method for minimizing uncertainty is given based on this model.The experimental results verify the validity of data filling and the extension of the characteristic relation model,and show that the scale with the least uncertainty can be selected as the optimal scale by the sequential three-way decision model.