改进的SIR模型在微博信息传播中的应用

摘要

研究微博中信息的传播规律对舆论预测与管控、市场营销等方面具有重要意义。当前的信息传播模型大多忽视了不同信息间、不同用户间的个体差异。为解决这一问题，本文选取了影响用户浏览和转发信息行为的特征，采用集成逻辑回归与SVM的二分类算法预测个体行为。预测多个用户对于同一信息的浏览与转发行为构成了本文中的信息传播模型。结果表明，该模型能较好地预测现实微博网络中的信息传播过程。%Studying diffusion rules of information in microblog is significant for public opinion predicting and controlling, marketing and etc. The most current information diffusion models ignore the diversity of different in-formation and different users. In order to solve the problem, this paper, by choosing the characteristics of users' browsing and retweeting behavior, uses binary-class classification algorithm based on logistic and SVM to predict individual behavior. In this paper, the information diffusion model is to predict users' browsing and retweeting be-havior for given information. Simulation result shows that this model can better predict the process of information diffusion in real microblog network.