基于有向渗流理论的关联微博转发网络信息传播研究

微博网络的快速性、爆发性和时效性, 以及用户复杂的行为模式, 使得研究其信息传播模型及影响因素成为网络舆情的热点方向. 利用压缩映射定理, 分析不动点迭代过程的收敛条件, 得到有向网络信息传播过程的渗流阈值和巨出向分支的数值解法; 通过可变同配系数生成模型, 分析关联特征对信息传播的影响; 最后利用微博转发网络数据进行仿真对比实验. 结果表明: 虽然四类关联特征同时体现出同配、异配特征, 但信息传播结果更多受入度-入度相关性、入度-出度相关性影响; 通过删除少量节点的方法, 提取边同配比例, 验证大部分节点的四类关联特征呈现一致性.

Information spreading in correlated microblog reposting network based on directed percolation theory

Due to the properties of rapidity, explosive, timeliness and complicated behavior for user, the research on information spreading progress and influence factors for microblog becomes a hot area of network public opinion. In this paper, firstly we use the contracting mapping principle to discuss the convergence conditions of the iterative algorithm. The numerical solution of the percolation threshold and the size of the largest out-component are proposed. Then the influence of assortativity is analyzed based on the generation model with varying parameter. The feasibility of the proposed algorithm is verified by collecting microblog reposting data. Experimental results demonstrate that four correlation characteristics are shown to have assortativity and disassortativity, but the results of message spreading are closer to that of the assortative network which is related to in-in and in-out degree correlation. It can be verified that the four types of correlation characteristics of a large part of nodes show their consistency for assortativity, through deleting a few nodes as well as extracting link scale for four degree correlations.