Research On Weibo Retweet Behavior Prediction And Communication Maximization

In the past ten years,as a new information media,online social networks have been fully developed because of its popularity and gradually many users.Well-known online social networks include Weibo and Tencent Weibo in China,and Facebook and Twitter in foreign countries.The users of these online social networks still serve as a node in the information diffusion in the virtual world different from the real world.Different from the real world,information diffusion in online social networks is realized through the user's retweet behavior,and users in this network are free to post their own speech.With the increasing number of users in this kind of social network,and the information updating speed is much faster than that of traditional paper media,social networks have gradually become the main way for people to obtain information,express themselves,and conduct information diffusion.The purpose of this paper is to build a model that predicts the user's retweet behavior by analyzing the posting history of users in Weibo,the user's network,the basic information of the user,and the text content of the target Weibo.Since information diffusion in online social networks has gradually become a hotbed for the occurrence and development of online public opinion,this paper studies the problem of maximizing the impact based on user retweet behavior.Specifically,the main work of this paper can be summarized as follows:(1)For most prediction models that ignore the relationship between the tweets posted before and after the user,a predictive model is proposed to predict the user's retweet behavior.Natural language processing technology is used to represent the Weibo text and the target Weibo text published by the user's history respectively,and network representation learning technology is used to represent the user's network.At the same time,considering that the basic information of the user is the inherent characteristics of the user itself,and to some extent,it also reflects the user's interest,this proposed model combines the user's history,the target Weibo,the user's network and the basic information of users to predict the retweet behavior of users.Then,the effectiveness of the prediction method proposed in this paper is verified.The results on the real,public and non-full network microblog data set show that the model can predict the results more accurately.(2)Considering that information diffusion in social networks is closely related to user retweet behavior,this paper proposes an independent cascade model based on users' retweet behavior.Because the user may receive the same information posted or retweeted by multiple neighbors at the same time,the calculation method of activation probability in this model is not only related to the propagation path and the retweet probability of users themselves,but also closely related to the selection of initial users.Therefore,the activation probability is estimated dynamically.(3)Aiming at influence maximization under the proposed propagation model,this paper presents an influence maximization algorithm based on greedy algorithm to find the node with the largest information propagation.On two real data sets of different sizes,the results of simulation experiments show that the influence range of the nodes found by the proposed algorithm is better than that of the common algorithms.