基于个体行为特征的时效网络中传播源定位研究

项目名称： 基于个体行为特征的时效网络中传播源定位研究

项目编号： No.61503110

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 自动化技术、计算机技术

项目作者： 周银座

作者单位： 杭州师范大学

项目金额： 20万元

中文摘要： 准确快速的定位传播源是舆情/疫情防控的关键。以往的研究多基于固定拓扑的静态网络，然而真实网络多为连边随着时间不断变化的时效网络，其时效特性对传播源定位产生重要影响并加大研究的难度。本项目基于实证数据，利用统计物理学的方法，围绕时效网络中的信息传播，着眼于时效结构与节点动力学之间的相互作用这一重要机制，致力于研究基于个体行为特性的时效网络中传播源定位问题。本项目利用n阶聚合网络近似理论及多层耦合网络分析方法研究基于时效结构-个体行为关联的传播动力学；发展并改进k-shell等算法挖掘时效网络中的关键节点；运用时效Dijkstra算法、最大似然估计及相关性分析最终定位传播源。本项目强调理论分析、算法设计和实证研究的有效结合，所得的结果将有助于更好地理解社会网络中的时效特征与人类社会行为之间的关系和内在机制，并以此为基础找到高效定位传播源的方法，为相关机构提供有效合理的舆情/疫情的防控方案。

中文关键词： 复杂网络；网络传播；时效网络；传播源定位；关键节点挖掘

英文摘要： Locating the source(s) quickly and accurately is the key to the prevention and controlling of social rumor/epidemic. Previous studies about source(s) location are mostly based on static networks with fixed topology. However, real networks are temporal ones whose edges are varying with time. This characteristics may play an important role in locating the diffusion source(s) and increase the difficulty to this problem. This project is aimed at locating source(s) in temporal networks based on individual behavior by using statistical methods with empirical data. The project will focus on information propagation of temporal networks, and pay close attention to the interaction between the temporal structure and node dynamics. The project will use n-order aggregate networks approximation theory and multilayer network analysis method to study propagation dynamics based on temporal structure-node dynamics interaction. The project will also develop the existing algorithms, such as improved k-shell algorithm, to mine the key nodes of temporal networks. Moreover, the project will use T-Dijkstra algorithm, the maximum likelihood estimation and correlation analysis to locate the diffusion source(s). The project will collectively utilize methods of theoretical analysis, algorithm design and empirical study, in order to provide better understanding in the relation and mechanism of the structure and dynamical behavior about temporal networks. The project will also provide efficient and accurate algorithms in locating the diffusion source(s), and offer effective measures to control rumor/epidemic.

英文关键词： complex networks;propagation dynamics;temporal networks;diffusion source(s) location;key node(s) mining