多粒度形式概念分析的类属性块优化  

Optimization of class-attribute block in multi-granularity formal concept analysis

作　　者：李金海 贺建君[1,2] 吴伟志 LI Jin-hai;HE Jian-jun;WU Wei-zhi(Data Science Research Center,Kunming University of Science and Technology,Kunming 650500,Yunnan,China;Faculty of Science,Kunming University of Science and Technology,Kunming 650500,Yunnan,China;School of Mathematics,Physics and Information Science,Zhejiang Ocean University,Zhoushan 316022,Zhejiang,China;Key Laboratory of Oceanographic Big Data Mining and Application of Zhejiang Province,Zhejiang Ocean University,Zhoushan 316022,Zhejiang,China)

机构地区：[1]昆明理工大学数据科学研究中心,云南昆明650500 [2]昆明理工大学理学院,云南昆明650500 [3]浙江海洋大学数理与信息学院,浙江舟山316022 [4]浙江海洋大学浙江省海洋大数据挖掘与应用重点实验室,浙江舟山316022

出　　处：《山东大学学报（理学版）》2020年第5期1-12,共12页Journal of Shandong University(Natural Science)

基　　金：国家自然科学基金资助项目(11971211,61976194);浙江省自然科学基金资助项目(LY18F030017)。

摘　　要：针对现有的多粒度形式概念分析的介粒度标记方法无法实现类属性块内部信息的跨粒度层组合,基于实际应用需要将类属性块内部信息进一步划分为子类,通过跨粒度层重新组合各子类以提出多粒度类属性块,在此基础上分析多粒度类属性块的内部结构,揭示决策蕴涵随多粒度类属性块粒度粗细变化进行更新的规律,完善了基于多粒度形式概念分析的多层次知识发现理论与方法。The existing meso-granularity labeled method in the multi-granularity formal concept analysis is not able to realize the cross-granularity combination for the internal information of class-attribute blocks. Motivated by this problem, this paper divides the internal information of a class-attribute block into sub-classes based on the requirements of practical applications, and puts forward multi-granularity class-attribute blocks by means of combining the obtained sub-classes in a cross-granularity way. Then it analyzes the internal structure of multi-granularity class-attribute blocks, and reveals the mechanism of how to update decision implications when the multi-granularity class-attribute block becomes from coarse to fine or fine to coarse. The obtained results in this paper can further improve the theory and method of multi-level knowledge discovery with multi-granularity formal concept analysis.

关 键 词：粒计算 形式概念分析 概念格 多粒度形式背景 多粒度类属性块 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]