Research On Information Spreading In Social Network Based On Human Dynamics:Empirical Analysis And Modeling

With the popularity of Web2.0and the fast development of related technologies, the Social Networking Service as a form of network application has developed rapidly. Nowadays, the social network has become an important platform for people to disseminate information, express views and interact with each other. Different from the traditional ways of information dissemination, user activity patterns and social network structure characteristics make the propagation of information becoming more complex and uncertain, and traditional mathematical model of spreading dynamics can’t depict the spreading phenomenon in online social network. In view of this, combined with the theories and methods of the disciplines of information science, human dynamics, complex network and spreading dynamics, we empirical analysis and modeling of information spreading in social network. In this thesis, we study the impact of human activity patterns on information propagation, structural characteristics of social networks and evolutionary mechanismes. Furthermore, we investigate the impact of the structural characteristics of social networks on the information propagation. Empirical results and the findings are consistent and have great value scientific research and application.The main contents and conclusions are as follows:(1) We study the impact of bursty human comment patterns on the popularity of online content. In this paper, we first analyses the dataset from Sina micro-blogging. The result suggests that users comment patterns have periodic and bursty. The popularity of micro-blogging (namely the number of comments of micro-blogging) diminishes over time with power-law. According to the users comment patterns, we establish the model of information propagation and give the theoretical analysis. The analytical result shows that the user comment patterns are closely related to micro-blogging popularity. Finally, we comparative analyze the theoretical results and empirical results, and they are all support the conclusion very well.(2) We investigate the impact of the temporal heterogeneity on the information propagation. Empirical studies have shown that human life rhythms and patterns of activity greatly affect the information dissemination, especially temporal heterogeneity. In addition, the prevalence of online social networking services makes users inundated with a lot of information. Based on these two factors we establish the model of information propagation. From the simulation results, we find that the propagation velocity increases monotonously with the increase of temporal heterogeneity exponent. Furthermore, the decay of propagation velocity, namely the newly infected individuals at each time step, also follows power law distribution. Meanwhile, the exponent characterizing the temporal heterogeneity is related to that in the decay of propagation velocity by the relationβ=α-1. These results are well supported by both the theoretical predictions and empirical data.(3) In order to understand the mechanism of the formation and evolution of social network, we propose the SD-CNN model. The model based on the mechanism of strength driven attachment and connecting the nearest neighbor, and reflects the real phenomenon of social network:two users who have common friends become friends is more likely than randomly selected two users. Simulation results show that the degree and strength of nodes follow power law distributions.(4) We investigate the impact of community structure on information propagation on social network. We established the propagation model not only consider the number of community structure, but also the number of links between the communities. The conclusions are as follows:for a piece of information propagation on social networks, we find that the number of links among communities determines the fraction of infected nodes. And the number of community structure also has great impact on information propagation process. The results can be useful for optimizing or controlling information, such as rumor or facts, propagate on online social networks.