Study Of User Influence Based On Micro-blog's Spreading Cascade And Its Application

In recent years,with the rapid development of social networks and the growing number of users,the influence of social networks is also growing.At the same time,social networks accelerated the dissemination of information,and become an important medium for information dissemination and diffusion.As a main online social networking platform,Micro-blog has many features,such as real-time publishing,real-time communication,multi-channel participation,simple and practical,and become an important part of people's daily life.In a Micro-blog network,there are special users who induce other users to actively utilize micro-blogs.Identifying such influential users is important when establishing business policy and business models for the Micro-blog network.This paper focus on the research of the Sina Micro-blog.First of all,this paper tries to defines the users whose micro-blogs exhibit significant influences over other users in a new way.These users are named super users.The new definition can divide(clustering)these super users from another point of view.Second,this paper constructs a model based on the dissemination of micro-blogs and the operation of users,this model can explain the diffusion of micro-blogs' s influence and can successfully capture the impact of the propagation and dissemination of the information in the Micro-blog network.Third,this paper proposes a method of identifying the super users,called User Influence Detection Algorithm Based on Micro-blog's Spreading Cascade,referred to as UISC Algorithm.In UISC,the influence of a blog is computed from both direct and spreading cascade ways.Then,the user influence is computed by adding the influence of all the micro-blogs owned by the user.Since a micro-blog with longer exposure tends to receive more user activities,the raw value of the influence of a micro-blog is inversely adjusted with exposure duration.What's more,this paper analyze the performance of the proposed method by applying it to an actual Micro-blog network and performed five sets of experiments.The first set of experiments were used to select the most “representative” parameter setting to be used in subsequent experiments.The second set of experiments compared the similarity between the power users selected by different methods.The third set of experiments compared the survival rates of the power users selected by different methods.The experimental results demonstrate that the definition ofsuper user is adequate to address the dynamic nature of the Micro-blogosphere and the main concerns of the Micro-blog industry.We also discuss the business models based on super users that could be used to stimulate user activities in a Micro-blog network.