Research On Information Filtering And Diffusion Mode Of Social Network

Social network has become an important carrier of information diffusion and attracted a lot of attention in various areas.A large number of research has been devoted to understanding the spreading process,including the construction of contagion models,empirical analysis,and the prediction of the diffusion.On the one hand,it is commonly believed that the human interactions play a major role in the process of information diffusion.It emphasizes that the individual decision process is determined by other individuals.However,in some actual scenes,the case is different.For example,in the process of forwarding a tweet on the network,people always pay more attention to the content of the tweet rather than the decision of other individuals.The influence from other individuals on the process of individual decision is weak.On the other hand,the information in reality may not be always true.Rumors may cause badly damage to the society.Therefore,some authoritative individuals are needed to distinguish the true and false information in time.We define these authoritative individuals as smart nodes who can filter the information.Due to the smart nodes,it is possible for the system to achieve such a state: the true news can spread widely throughout the network while the false ones can be quickly suppressed,which verifies the old saying---"rumor stops in wise man".Based on the information filtering model,this paper mainly considers the following two problems: the influence of the network structure on information diffusion and the selection strategies of smart nodes.We firstly put forward the information diffusion model based on the true and false message and analyzed the spreading dynamics based on the mean-field theory.We find a critical threshold condition in the propagation process.Then we compare the influence of several classic structures on the spreading dynamics.It is found that the assortative network with “chain” structure is more difficult to spread information than the disassortative network with “star” structure.Finally,we propose different selection strategies of smart nodes for different network structures.Our model aims to provide some insights and reference for the information diffusion research.