代数连通性在社会网络影响力传播最大化中的应用研究  

Application research of algebraic connectivity in influence propagation maximization of social network

作　　者：赵富强[1] 杨贵军 王双琳 何丽[1] Zhao Fuqiang;Yang Guijun;Wang Shuanglin;He Li(Dept.of Information Science&Technology,College of Science&Technology,Tianjin University of Finance&Economics,Tianjin 300222,China)

机构地区：[1]天津财经大学理工学院信息科学与技术系,天津300222

出　　处：《计算机应用研究》2018年第1期177-181,共5页Application Research of Computers

基　　金：国家自然科学基金资助项目(11471239);天津自然科学基金资助项目(15JCYBJC16000);天津市哲学社会科学研究规划基金资助项目(TJTJ15-002)

摘　　要：针对社会网络信息传播的研究多集中于具体应用场景中的传播机制,而用户间关系所形成的网络拓扑结构对传播的影响却没有给予充分考虑,因此对信息扩散的分析应当综合考虑社会网络的传播机制和结构特征,把相关系数引入到割边模型可以有效解决重叠社团划分问题。在考虑社会网络拓扑结构特征的基础上,提出了基于代数连通性的社会网络影响传播最大化模型。模型以代数连通性为主要参量计算边的中心性,实现网络社区的快速划分,通过降维达到算法效率优化;模型挖掘社区内影响力大的核心节点、度中心节点和社区间的桥节点,从中选择top-k全局影响力最大的节点作为初始传播点集。实验结果显示,模型在传播影响范围和运行时间上都更具优势。The recent researches of information propagation in social networks focus on the application of the spreading met-hods,which don’t consider the influence of network topologies formed by the relationship between users.Therefore,the analysis of information diffusion should study propagation mechanism and structure feature for the social networks.Cut model with correlation coefficients could resolve overlapping community detection by minimizing the algebraic connectivity of complex networks.In the view of the social network topology feature,this paper proposed the influence propagation maximization model in social network based on algebraic connectivity.The model calculated the measure of edge centrality based on algebraic connectivity,cut the community rapidly and realized the goal of optimization algorithm efficiency by dimension reduction.The model firstly searched for regional influence core nodes and degree central nodes in the community.Secondly,it detected bridge nodes between the communities by the second eigenvector of Laplacian and chose the top-k nodes of global influence from the three nodes set as the initial spreading nodes set.The experiment results show that the model is more advantages in propagation influence range and running time.

关 键 词：社会网络 影响传播最大化 相关系数 代数连通性 社区发现 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]