摘要
数据分析中,从网络中进行概念认知学习是网络背景下的机器学习或人工智能领域的重要问题.首先通过分析复杂网络方法与形式概念方法的数据基础,将二者的数据通过邻接矩阵与关联矩阵统一起来,提出一种网络形式背景框架,使以上两种理论与方法之间有了互通的桥梁,从而可以结合它们各自的优势对网络概念进行更深入的研究.在此基础上,从网络概念的三个层次出发研究了以下内容:(1)通过定义节点的结构影响力和内涵影响力并将它们进行加权,定义了节点的网络影响力.(2)通过分析扩散网络、收缩网络的特点提出强概念、弱概念、网络概念,并给出了网络概念的特征值:概念的势、概念平均度.于是,该理论不仅能在网络中找到网络概念,还能给出网络概念的重要性和网络概念内部的差异性.(3)研究了强(弱)概念的有关性质,为以后构造相应的代数系统,生成各种网络概念算子提供了理论基础。
In data analysis,concept . cognitive learning from the network is an important issue in the field of machine learning or artificial intelligence in the context of network. Firstly,by analyzing the data foundation of complex network and formal concept methods,the data induced by them are unified through the adjacency matrix and the association matrix,and a formal context of network framework is proposed. It is a bridge between the above two theories and methods,so that the research of network concept will be more deeply and meaningful by combining their respective advantages. On this basis,the following contents are studied from three hierarchies of the network concept:(1)by defining the structural influence and connotation influence of the nodes and weighting them,the network influence of the nodes is defined.(2)By analyzing the characteristics of diffusion network and convergent network,strong concept,weak concept,network concept are defined,meanwhile the eigenvalues of network concept: concept average diversity,concept average degree are given. Therefore,not only can the network concept be found in the network by the theory,but also the importance of the network concept and the internal differences of the network concept will be given by the theory.(3) The related properties of the strong (weak) concept are studied,which provides a theoretical basis for constructing the corresponding algebraic systems and generating various network concept operators.
作者
马娜
范敏
李金海
Ma Na;Fan Min;Li Jinhai(Faculty of Science,Kunming University of Science and Technology,Kunming,650500,China;Data Science Research Center,Kunming University of Science and Technology,Kunming,650500,China)
出处
《南京大学学报（自然科学版）》
CAS
CSCD
北大核心
2019年第4期609-623,共15页
Journal of Nanjing University（Natural Science）
基金
国家自然科学基金(61562050,61573173,61573321,41631179)
关键词
复杂网络
形式概念
网络形式背景
网络影响力
网络概念
complex network
formal concept
network formal context
network influence
network concept
