面向属性变化的动态邻域粗糙集知识更新方法  

Dynamic neighborhood rough sets approaches for updating knowledge while attributes ggeneralization

作　　者：胡成祥[1] 张莉[2] 黄晓玲[1] 王汇彬[1] HU Chengxiang;ZHANG Li;HUANG Xiaoling;WANG Huibin(School of Computer and Information Engineering,Chuzhou University,Chuzhou 239000,Anhui,China;School of Computer Science and Technology,Soochow University,Suzhou 215006,Jiangsu,China)

机构地区：[1]滁州学院计算机与信息工程学院,安徽滁州239000 [2]苏州大学计算机科学与技术学院,江苏苏州215006

出　　处：《山东大学学报（理学版）》2023年第7期37-51,共15页Journal of Shandong University(Natural Science)

基　　金：安徽省自然科学基金资助项目(2008085MF224,2108085MF215);江苏省高校自然科学研究项目(19KJA550002);江苏省六大人才高峰项目(XYDXX-054);江苏高校优势学科建设工程项目;安徽省教育厅高校自然科学研究重大项目(KJ2020ZD63);安徽省教育厅高校自然科学研究重点项目(2022AH051099,KJ2020A0726,KJ2021A1065,KJ2021B03);安徽省高校优秀青年人才支持计划项目(gxyq2022098)。

摘　　要：提出了属性集变化时动态更新三支决策区域的矩阵方法。首先,利用邻域距离矩阵和邻域关系矩阵等构建了面向决策类的邻域三支决策区域的表示;其次,考虑邻域决策系统中的属性变化情况,分析邻域关系矩阵等相关矩阵的更新策略;然后利用相关矩阵的更新策略,分别提出了属性增加和属性删除时更新三支决策区域的矩阵方法;最后,通过公共数据集验证了基于矩阵的增量更新方法的有效性。We propose the matrix-based dynamic methods for updating three-way regions in neighborhood decision systems while attributes generalizing.First,we construct the matrix representation strategy of three-way regions for each decision class by means of neighborhood relation matrix and neighborhood distance matrix.Subsequently,considering the variation of the attributes,the updating strategies for related matrices are analyzed by using prior knowledge.Based on these strategies,we propose the matrix-based incremental methods for updating neighborhood three-way regions of each decision class.Finally,we evaluate the efficiency of the proposed incremental methods on public available data sets.

关 键 词：邻域粗糙集 动态更新 三支决策区域 矩阵 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]