面向不完备混合数据的矩阵增量知识维护方法研究  

Matrix-based Incremental Approaches to Maintaining Knowledge for Incomplete Hybrid Data

作　　者：黄倩倩[1,2] 李天瑞[1,2] 杨新 王国强[1,2] 胡节 HUANG Qian-qian;LI Tian-rui;YANG Xin;WANG Guo-qing;HU Jie(College of Computer Science and Technology,Southwest Jiaotong University,Chengdu 611756,China;Institute of Artificial Intelligence,Southwest Jiaotong University,Chengdu 611756,China)

机构地区：[1]西南交通大学计算机科学与技术学院,成都611756 [2]西南交通大学人工智能研究院,成都611756

出　　处：《小型微型计算机系统》2020年第4期868-877,共10页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(61573292,61572406,61603313)资助;中央高校基本科研业务费项目(2682017CX097)资助。

摘　　要：现有邻域粗糙集模型可用于处理包含名义型和数值型两种类型共存的混合数据,但较少考虑混合数据的不完备性.本文从缺失值的两种语义解释出发,即"不关心值"和"丢失值",通过定义邻域特征关系和量化邻域特征关系,提出了面向不完备混合数据的两种新型邻域粗糙集模型,并给出了粗糙邻域近似知识的矩阵计算表达方法.此外,在属性集动态变化下,介绍了基于扩展邻域粗糙集模型的快速增量知识维护机理和方法.最后通过实例验证了所提出增量更新方法的有效性.The existing neighborhood rough set models are employed to process the hybrid data containing both categorical and numerical types,but the incompleteness of hybrid data is rarely taken into consideration.Therefore,starting from two different semantic explanations of unknow n values,i.e.,the"do not care"unknow n values and"lost"unknow n values,two extended neighborhood rough set models are proposed to handle incomplete hybrid data based on two novel binary relations,i.e.,the neighborhood characteristic relation and the neighborhood valued characteristic relation.Then,the matrix-based methods for computing a pair of approximations are presented.M oreover,the matrix-based incremental mechanisms for maintaining know ledge are introduced under attribute generalization.Finally,an illustration is employed to validate the effectiveness of the proposed incremental methods.

关 键 词：不完备混合数据 粗糙集 邻域关系 矩阵运算 增量更新 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]