Improvement And Validation Of Influence Propagation Model In Microblog

As the most convenient and most efficient information dissemination social networking platform, microblog has penetrated people’s daily lives. Such as ordinary Internet users, public figures, the media, government officials and enterprises, people are more and more accustomed to publish or disseminate the latest news through the microblog. With the rapid development of the research in social network, microblog as a typical social network, also attracted wide attention from scholars. In this paper, we begin on the influence spread, mainly focused on the influence propagation model in microblog.Influence spread in microblog is mainly reflected by the dissemination of information, so the analysis of factors affecting the information diffusion is of great significance to improve the influence propagation model. In the influence propagation model, factors affecting dissemination of information are the factors affecting node activation probability. Through in-depth study and analysis of existing models of the factors affecting the dissemination of information in social networks, we propose three factors that affecting node activation probability in microblog, that is, the tweet feature, the relationship feature and the user feature. We provide the corresponding calculation method, which is called TRU (Tweets, Relationships and Users) node activation probability calculation methods. The tweet feature mainly measures the tweets’ degree of influence in the spread. The tweet is divided into four kinds of forms, that is, the plain text tweets, the tweets only containing URL, the tweets only containing multimedia and the tweets containing both URL and multimedia. The relationship feature depends on interactions and similarities between the two users, say, behavioral inertia between them and similarity of their interests. The user feature mainly compute the global influence of each user in microblog. Combining with the TRU node activation probability calculation method and the improved node activation threshold calculation method, the paper proposes the ICTRU (Independent Cascade model with TRU measurement) influence propagation model based on the Independent Cascade model.In the paper, we collected more than4.5million users and its corresponding tweet information to verify the validity of ICTRU model through application programming interface provided by "weibo open platform". By contrast ICTRU model’s propagation figure and the actual propagation figure of the original tweet, simultaneous analysis of both the corresponding tree data, it is confirmed that ICTRU model can effectively simulate the influence propagation in microblog. Meanwhile, by contrast ICTRU model and the original Independent Cascade model, it is proved that ICTRU model can better apply to influence spread in microblog.