Empirical Study Of Weibo Spreading Based On Complex Social Networks

Weibo is an open platform for users to report, share and get information based on relations in the era of Web2.0. Depending on great number of users and the’back to face’type of weak relations among them, the spreading of weibos is very fast, which is similar to the diffusion of virus and with the power like nucleus fission. Repost is an important message propagation method on the Weibo platform. In this article, the Weibo reposting user network is researched from the angle of complex social network. The methodology of social network analysis and network diffusion dynamics is utilized to discuss the network topology features and spreading characteristics of common weibos’reposting user networks. The aim of this research is to understand the impact of user interaction to information spreading and also find the information propagation rule on the open platform. It is helpful to the study of information spreading in social networks. Main work and conclusion are summarized as follows.First, research data is collected from Sina Weibo platform utilizing self-developed program. Then, the Weibo reposting user network model is established through social network analysis methodology. According to the measurement and comparison of topology eigenvalues including clustering coefficient, average path length and degree distribution of those network models, it is verified that the weibo reposting user network also has small world and power-law distribution characteristics and therefore belongs to complex social networks.Second, the correlativity between network topology eigenvalues and the number of weibo repost is tested by statistical information processing technology. It is found that the network average degree has significant negative correlation with and the number of weibo repost; the average path length has significant positive correlation with the number of weibo repost; the clustering coefficient has significant positive correlation with the number of weibo repost; and the subgroup analysis value has significant positive correlation with the number of weibo repost. What’s more, the intrinsic reasons why those correlations may appear are analyzed from the angles of network evolution, the possibility of a person seeing a message, neighbor effect, and common attributes among users.Finally, the Weibo repost dynamics model is constructed through broadening the assumptions of the traditional SI model. Based on the real data collected from Sina Weibo platform, the rate of infection λ and the birth rate μ. are estimated. It is found that a weibo is propagated fast when it is firstly posted. As for some weibos with relatively big amount of repost and long period of reposting, though the absolute number of repost increases with the join of hubs, the infection density decreases because of the dynamic growth of the reposting user network.