A Study Of Information Diffusion In Complex Networks

Information diffusion has become a hot topic in interdisciplinary research. With the explosive growth of information available in Web2.0 era, surfers publish, share and disseminate information using social media such as Microblog and We Chat. While information diffusion bring unlimited business opportunities for the promotion of new idea,new technologies and new products, it also causes concern about its potential hazard to social stability, and may even lead to social unrest. By studying the models and rules of information diffusion, it’s possible to control rumors, monitor public opinion and guide information. Taking into account its internal and external factors, we conduct experimental modeling and analysis on information diffusion at macro-, meso- and micro- levels based on the complex network, which holds great theoretical and practical significance.Our contributions are as follows:1. Propose a new approach to identify influence spreader in complex network—KSCmodel. We analyze the deficiencies of classical evaluation metrics including degreecentrality, betweenness centrality, closeness centrality and K-shell centrality. Wealso propose a method where the influence spread is determined by internal andexternal factors. The SIR model is used to simulate the propagation process in fourreal complex networks. The experimental results demonstrate that our KSC modelbased method can select the most influential spreaders with higher precision andbroader scope of application.2. Propose a new approach for influence maximization in complex network—RMDNmodel. Influence maximization is a combinatorial optimization problem of findingthe most influential spreaders, which is shown to be NP hard. We find the seednodes of propagating source with the local Information of randomly selected nodesand their directly connected neighbor nodes. We also give the theoretical deriva-tion and analysis of our algorithm, and we prove its feasibility. The experimentalresults of our RMDN model are similar and sometimes slightly better than classi-cal algorithms in four real complex network, while the time complexity is reducedsignificantly.3. Propose an information spread structural diversity model—ISSD model. We an-alyze both the temporal variations of information diffusion and the internal andexternal factors of spreaders. We propose five assumptions of information diffusion, analyze the internal attributes of individuals in the propagation process, and consider the structural diversity of a user’s Ego network as well as the external environmental factors. Both the information diffusion mechanism and the information diffusion based on ISSD model are formalized in this paper. Our experimental analysis shows how time and structural diversity affect the information diffusion. The modeling and quantitative analysis also help in better understanding the process of information diffusion.