ISSN 1000-1239 CN 11-1777/TP

计算机研究与发展 ›› 2020, Vol. 57 ›› Issue (2): 447-458.doi: 10.7544/issn1000-1239.2020.20190279

• 人工智能 • 上一篇    

多粒度形式概念分析的介粒度标记方法

李金海1,2, 李玉斐1,2, 米允龙3, 吴伟志4,5   

  1. 1(昆明理工大学数据科学研究中心 昆明 650500);2(昆明理工大学理学院 昆明 650500);3(中国科学院大学计算机与控制学院 北京 100190);4(浙江海洋大学数理与信息学院 浙江舟山 316022);5(浙江省海洋大数据挖掘与应用重点实验室(浙江海洋大学) 浙江舟山 316022) (jhlixjtu@163.com)
  • 出版日期: 2020-02-01
  • 基金资助: 
    国家自然科学基金项目(11971211,61976194,61562050,61573173,61573321,41631179)

Meso-Granularity Labeled Method for Multi-Granularity Formal Concept Analysis

Li Jinhai1,2, Li Yufei1,2, Mi Yunlong3, and Wu Weizhi4,5   

  1. 1(Data Science Research Center, Kunming University of Science and Technology, Kunming 650500);2(Faculty of Science, Kunming University of Science and Technology, Kunming 650500);3(School of Computer and Control Engineering, University of Chinese Academy of Sciences, Beijing 100190);4(School of Mathematics, Physics and Information Science, Zhejiang Ocean University, Zhoushan, Zhejiang 316022);5(Key Laboratory of Oceanographic Big Data Mining and Application of Zhejiang Province (Zhejiang Ocean University), Zhoushan, Zhejiang 316022)
  • Online: 2020-02-01
  • Supported by: 
    This work was supported by the National Natural Science Foundation of China (11971211, 61976194, 61562050, 61573173, 61573321, 41631179).

摘要: 现有的多粒度标记形式背景将所有属性的粒度标记值通过多个单粒度标记形式背景的并置予以表示,这可能导致后续相关问题的研究主要集中于多个单粒度标记形式背景的知识发现及其相互关系,不利于挖掘多粒度标记形式背景的多层次知识.通过重构单粒度标记形式背景的属性粒度标记值,在多粒度标记形式背景中研究介粒度标记形式背景,使得知识发现不仅仅局限于数据采集或表示所形成的粗细粒度标记数据,还包括跨粒度重组数据的知识发现.首先,提出介粒度标记形式背景的概念,给出相应的语义解释;其次,讨论介粒度标记形式背景的泛化与特化,在此基础上证明所有介粒度标记形式背景在泛化-特化关系下构成一个完备格;然后,给出多粒度标记决策形式背景的介粒度知识发现方法,得到粗细介粒度标记形式背景之间的决策蕴涵诱导关系;最后,实验分析表明介粒度标记方法的有效性和优势所在.

关键词: 粒计算, 形式概念分析, 粗糙集, 概念格, 决策形式背景

Abstract: For the existing multi-granularity labeled formal contexts, the granular labeled values of all the attributes are organized by the union of some single-granularity labeled formal contexts. This may lead to the result that the subsequent related research problems mainly concern knowledge discovery on these single-granularity labeled formal contexts as well as their internal relationships. Consequently, it will not be beneficial for mining multilayer knowledge from multi-granularity labeled formal contexts. In this paper, we discuss meso-granularity labeled formal contexts in multi-granularity labeled formal contexts by restructuring granular labeled values of attributes of the original single-granularity labeled formal contexts, which makes knowledge discovery not limited to coarse and fine granular labeled data formed by data acquisition or representation, but absorbed from the combined data through cross-granularity. Firstly, we give the notion of a meso-granularity labeled formal context and its corresponding semantic interpretation. Secondly, we investigate generalization and specialization of meso-granularity labeled formal contexts, and prove that all the meso-granularity labeled formal contexts form a complete lattice under the generalization-specialization relation. Thirdly, we put forward a meso-granularity based knowledge discovery method for multi-granularity labeled formal decision contexts, and clarify the inference relationship between decision implications extracted from the coarse and fine meso-granularity labeled formal contexts. Finally, our experimental analysis demonstrates the effectiveness and advantages of the proposed meso-granularity labeled methods.

Key words: granular computing, formal concept analysis, rough set, concept lattice, formal decision context

中图分类号: