计算机工程与应用 ›› 2021, Vol. 57 ›› Issue (1): 84-91.DOI: 10.3778/j.issn.1002-8331.2002-0368

• 大数据与云计算 • 上一篇    下一篇

以概念格为背景的关联规则可视化

杨葛英,沈夏炯,史先进,张磊   

  1. 1.河南大学 计算机与信息工程学院,河南 开封 475004
    2.河南大学 河南省大数据分析与处理重点实验室,河南 开封 475004
  • 出版日期:2021-01-01 发布日期:2020-12-31

Visualization of Association Rules in Context of Concept Lattices

YANG Geying, SHEN Xiajiong, SHI Xianjin, ZHANG Lei   

  1. 1.School of Computer and Information Engineering, Henan University, Kaifeng, Henan 475004, China
    2.Henan Key Laboratory of Big Data Analysis and Processing, Kaifeng, Henan 475004, China
  • Online:2021-01-01 Published:2020-12-31

摘要:

传统的关联规则表示方法无法展示概念之间的本质关系,缺少对概念层面的认识,忽略了知识发现结果的共享等问题,而概念格作为一种能够生动简洁地体现概念之间泛化和例化关系的数据结构,在对关联规则可视化和发现潜在知识方面也有着独特的优势。提出了以概念格为背景的关联规则可视化方法,以概念为查找单元,在概念格中寻找需要展示的关联规则路径,将属性之间的关联关系扩展到概念层面,并给出了相对应的多模式规则的可视化的策略与算法。结合某校图书馆借书记录数据,进行关联规则分析与可视化实现。实验结果表明,该可视化方法在知识发现和共享方面具有良好的效果。

关键词: 概念格, 关联规则, 泛化和例化, 知识发现, 可视化

Abstract:

Traditional rule representation methods cannot show the nature of the relationship between concepts, the lack of understanding of the concept hierarchy, ignoring the problem such as sharing the results of knowledge discovery, and concept lattice as a data structure can succinctly vivid embodiment of generalization and instantiated in the relationship between the concepts, in terms of knowledge visualization and association rules found potential also has a unique advantage. This paper proposes an association rule visualization method with concept lattice as the background, takes concept lattice as the search unit, looks for the association rule path to be displayed in concept lattice, extends the association relation between attributes to the conceptual level, and gives the corresponding visualization strategy and algorithm of multi-pattern rules. Finally, the association rules are analyzed and visualized based on the library data. Experimental results show that the visualization method is effective in knowledge discovery and sharing.

Key words: concept lattice, association rules, generalization and instantiated, knowledge discovery, visualization