基于不完备混合序信息系统的增量式属性约简

由于大数据环境下数据呈现出动态更新的特征,因此增量式属性约简已成为粗糙集理论的重点研究方向。本文针对不完备混合型有序信息系统,利用邻域优势条件熵提出一种对象更新情形下的增量式属性约简算法。首先,针对不完备混合型有序信息系统提出一种新的邻域优势粗糙集模型,同时在其基础上定义了邻域优势条件熵,并设计出一种不完备混合型有序信息系统的非增量式属性约简算法；然后,针对不完备混合型有序信息系统对象的动态性,分别研究了邻域优势条件熵随信息系统对象增加和对象减少时的增量式更新；最后,利用邻域优势条件熵作为启发式函数提出了不完备混合型有序信息系统对象增加和对象减少时属性约简的增量式更新算法。实验结果表明,所提出的增量式算法无论在属性约简结果和属性约简效率上均比非增量式算法具有更高的性能。

Incremental Attribute Reduction Algorithm Based on Incomplete Hybrid Order Information System

Because the data in the big data environment presents the characteristics of dynamic updating, incremental attribute reduction has become the key research direction of rough set theory. In this paper, an incremental attribute reduction algorithm in the case of object update is proposed by using neighborhood dominance conditional entropy for incomplete hybrid ordered information system. Firstly, a new neighborhood dominance rough set model is proposed for incomplete hybrid ordered information system. Based on it, the neighborhood dominance conditional entropy is defined, and a non-incremental attribute reduction algorithm for incomplete hybrid ordered information system is designed; Then, aiming at the dynamics of objects in incomplete hybrid ordered information system, the incremental update of neighborhood dominance conditional entropy in the environment of object increase and object decrease is studied respectively; Finally, the incremental update of neighborhood dominance conditional entropy is used to design incremental attribute reduction algorithms for object increase and object decrease in incomplete hybrid ordered information system. Experimental results show that the proposed incremental algorithm has higher performance than the non-incremental algorithm in attribute reduction results and attribute reduction efficiency.