决策连续形式背景的可视化数据离散化方法  

Visual discretization for decision continuous formal context

作　　者：张涛[1] 师浩斌 李林[1] 李朝辉[1] 

机构地区：[1]燕山大学信息科学与工程学院,河北秦皇岛066004

出　　处：《计算机应用研究》2016年第2期388-391,395,共5页Application Research of Computers

基　　金：国家自然科学基金资助项目(61273019);河北省自然科学基金资助项目(F2015203013);河北省社会科学基金年度项目(HB14YY005);国家社会科学基金资助项目(12BYY121)

摘　　要：连续形式背景离散化是形式概念分析领域重要的基础问题之一。针对形式背景离散化的特殊要求,提出了一种可视化的数据离散化方法。该方法借助可视化方法对数据类别分布进行表示,将连续数据分布转换为图形分布,进一步利用视觉模糊性对图形空间进行处理,进而将决策连续背景离散化。通过UCI数据集上的实验表明,与传统离散化方法相比,采用该方法进行数据离散化后的二值形式背景具有结构简单且不失准确性的优点。Discretization for decision continuous formal context is one of basic issues in the field of formal concept analysis. According to the special requirements of formal concept analysis field for data discretization, this paper presented a method which is in the form of visualization. This method presented the data distribution by visualization, converted the data distribu- tion to figure distribution, and fuzzy analysis the figure to discretize the decision continuous formal context. The whole progress considered not only the category distribution, but also the fuzzy of the data. The experiments show that compared to the tradi- tional discretization methods, the binary formal context after using this method to diseretize data has the advantages of simple structure and without loss of accuracy.

关 键 词：形式背景 离散化 可视化 形式概念分析 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]