模糊语言属性偏序结构图的逐层属性约简算法  

A Layer-by-Layer Attribute Reduction Algorithm for Fuzzy Linguistic Attribute Partial Order Structure Diagram

作　　者：庞阔 周爱 杨鑫冉 李楠[2] 邹丽 鲁明羽[1] PANG Kuo;ZHOU Ai;YANG Xinran;LI Nan;ZOU Li;LU Mingyu(Information Science and Technology College,Dalian Maritime University,Dalian 116026;School of Computer and Information Technology,Liaoning Normal University,Dalian 116081;School of Computer Science and Technology,Shandong Jianzhu University,Jinan 250102)

机构地区：[1]大连海事大学信息科学与技术学院,大连116026 [2]辽宁师范大学计算机与信息技术学院,大连116081 [3]山东建筑大学计算机科学与技术学院,济南250102

出　　处：《模式识别与人工智能》2022年第9期774-788,共15页Pattern Recognition and Artificial Intelligence

基　　金：国家自然科学基金项目(No.61976124)资助。

摘　　要：在形式概念分析中,属性偏序结构图作为一种数据可视化工具,可有效解决用户认知过载的问题.在现实生活中,人们往往通过模糊语言值表达偏好信息,会产生大量的模糊语言值数据.为了解决在模糊语言环境下的属性约简问题,文中提出模糊语言属性偏序结构图的逐层属性约简算法.首先,基于模糊语言值形式背景构建模糊语言属性偏序结构图,将模糊语言值数据嵌入属性偏序结构图中.通过语言真值格蕴涵代数作为模糊语言值表示模型,表达模糊语言值间的序关系和不可比关系.然后,为了获取保持模糊语言值形式背景区分能力不变的最小属性子集,结合模糊语言值属性偏序结构图,搜索未与底层节点建立边的节点.在保证模糊语言值属性偏序结构图类等价的前提下,计算该节点及其子节点间的差别属性,并构造相应的逐层属性约简模型.最后,通过实例与对比实验验证文中算法的有效性和实用性.As a data visualization tool,attribute partial order structure diagram can solve the problem of user cognitive overload in formal concept analysis effectively.People often express preference information through fuzzy linguistic values in real life,and thus a large amount of fuzzy linguistic-valued data is generated.To solve the problem of attribute reduction in fuzzy linguistic environment,a layer-by-layer attribute reduction algorithm for fuzzy linguistic attribute partial order structure diagram is proposed in this paper.Firstly,the fuzzy linguistic attribute partial order structure diagram is constructed based on the fuzzy linguistic-valued formal context,and the fuzzy linguistic-valued data is embedded into the attribute partial order structure diagram.The order relation and the incomparable relation between fuzzy linguistic values are expressed by the linguistic truth-valued lattice implication algebra acting as the representation model of the fuzzy linguistic values.Secondly,the fuzzy linguistic attribute partial order structure diagram is employed and the nodes that do not form edges with the underlying nodes are searched to obtain the minimum attribute subset with the fuzzy linguistic-valued formal context discrimination ability unchanged.On the premise of ensuring the class equivalence of the fuzzy linguistic attribute partial order structure diagram,the difference attribute between the node and its child nodes is calculated,and the corresponding layer-by-layer attribute reduction model is constructed.Finally,examples and comparative experiments verify the effectiveness and practicability of the proposed method.

关 键 词：模糊语言属性偏序结构图 属性约简 模糊语言值形式背景 语言真值格蕴涵代数 模糊语言值分层概念格 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]