Social Network Maximizing Influence Model And Computing Method Research

With the rapid development and wide application of social networks,platforms such as Facebook,Wechat,Weibo bring huge amounts of network traffic and user data,which lead to the rich research of social marketing.In the field of social marketing,maximizing influence is one of its key research directions.Traditional research focuses on designing and optimizing information diffusion models for calculating the influence of nodes in social networks,and improving the corresponding algorithms to quantify the influence value of nodes in the models,which positively promotes the research of social networks.However,when using social networks for product marketing,companies often tend to face a variety of practical factors to propagate product information,how to solve the effects of multiple factors is an valuable work.This article focuses on maximizing influence to search seed users to disseminate product information,and also takes into account the user's interest preference to insure that the target users can be influenced by seed users are interested in the product;and companies also need to control the marketing cost to obtain greater marginal revenue.Based on the above discussion,this paper models influence maximization as a multiobjective optimization problem by considering practical features such as diffusion scale,user interest preference and diffusion budget,proposes the multiobjective influence maximization(MOIM)model.In order to solve the NP-hard problem of influence maximization,we use Monte Carlo sampling method to obtain influential users.Then a seed selection algorithm of multiobjective evolutionary algorithm(MOEA/D)based on decomposition strategy is proposed to combine and optimize seed to solve MOIM model.We use real social network data to verify the performance of the model and method.However,its calculation process is time-consuming,we use the distributed framework Map Reduce to parallelize the process to speed up the calculation.Our experiments show that the proposed model can generate proper seed sets to satisfy different purposes of information propagation.Sensitivity analysis proves that our model is robust under various practical conditions.