属性集变化下序决策信息系统的增量属性约简算法

Incremental attribute reduction algorithm for ordered decision information systems with the change of attribute set

当序决策信息系统中的属性集不断变化时,基于优势关系的现有静态算法无法高效地更新其属性约简,为此,从属性增加和属性删除两个角度出发,以知识粒度表征的属性重要度为启发信息,提出两种新的增量属性约简算法.首先介绍优势粗糙集方法的相关基础知识,并将经典粗糙集中基于知识粒度的属性约简算法扩展到优势粗糙集方法,得到可处理序决策信息系统的属性约简算法;然后,给出劣势属性矩阵的定义,并基于知识粒度的矩阵计算方法分析属性增删时属性约简的增量式更新机制,进一步设计了两种增量属性约简算法.最后,分析比较三种算法的时间复杂度,选取了六个不同的UCI数据集进行算法性能的测试,结果表明,提出的算法比静态的属性约简算法更高效.

When attribute set in the ordered decision information system is constantly changing,existing static algorithms that have been studied based on dominance relationship cannot efficiently update its attribute reduction.To this end,this paper proposes two new incremental attribute reduction algorithms from the perspective of both attribute addition and attribute deletion,respectively,using the attribute importance of knowledge granularity representations as the heuristic information.Firstly,the relevant basic knowledge of the dominance rough set method are introduced,and the attribute reduction algorithm based on knowledge granularity in the classical rough set is extended to the dominance rough set method to obtain an attribute reduction algorithm that can handle ordered decision information systems;Then,the definition of the inferior attribute matrix is given,and the incremental update mechanism of attribute reduction during attribute addition and deletion is analyzed by the matrix calculation method of knowledge granularity.From there,two incremental attribute reduction algorithms are further designed;Finally,time complexity of the three algorithms is analyzed and compared,and six different UCI datasets are selected to test the algorithm performance.The test results show that the algorithm proposed in this paper is more efficient than the static attribute reduction algorithm.