大数据偏序结构生成原理  

Generation principle of partial ordered structure towards big data

作　　者：洪文学[1,2] 李少雄[1] 张涛[3] 栾景民[1] 刘文远[3] 

机构地区：[1]燕山大学电气工程学院,河北秦皇岛066004 [2]东北大学秦皇岛分校大数据可视化分析技术中心,河北秦皇岛066004 [3]燕山大学信息科学与工程学院,河北秦皇岛066004

出　　处：《燕山大学学报》2014年第5期388-393,402,共7页Journal of Yanshan University

基　　金：国家自然科学基金资助项目(61273019;61201111;81273740;81373767);河北省自然科学基金资助项目(F2013203368)

摘　　要：形式概念分析是一种强有力的数据分析和可视化工具,自提出至今已被应用到数据挖掘、知识发现等领域。但由于概念间复杂的关系使得概念格中的连线错综复杂,尤其是在处理大的形式背景时显得尤为混乱。形式背景中所包含的属性间关系、对象间关系以及对象-属性关系是形式背景的本质关系,本文在人类认知事物哲学原理的指导下,构建了以描述属性间关系和区分对象为基本目的的偏序结构图,并描述了其构建方法。该种图形层次分明、结构清晰,不存在线的交叉连接,具有比较好的可视化效果,尤其计算方法简单,存在应用到大数据的潜能,为大数据数据分析和知识发现提供了一种潜在工具。Formal concept analysis is a powerful tool in data analysis and visualization, and has been applied to data mining,knowledge discovery andmany other fields since proposed. However, in the concept lattice, the complex relations between conceptsmake the lines rather complicated and crossed, especially when dealing with a large-scale formal context. The relation among attributes,objects and attribute-object are the essential relations in a formal context. Therefore, under the guidance in the philosophicalprinciple of human being＇s cognition, the partial ordered structure diagram aiming to delineate the relations among attributes anddistinguish distinctive objects is proposed, and construction method is described. Its distinct hierarchy, clear structure, uncrossedlines provide a better visualization. Apart from that, simple computational method of it makes a large potential in allusion to bigdata. Hence, a novel and efficient tool towards data mining and knowledge discovery of big data is provided by this diagram.

关 键 词：形式背景 偏序结构 大数据 属性偏序结构图 对象偏序结构图 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]