The Key Technology Research Of Information Mining And Propagation Based On Microblog Networks

Microblog belongs to a new service of online social networks. In the past few years this Internet application which has strong "self-media" properties has already been an important social media service due to its weak link, open, convenient and real-time features. It may cause profound impacts on the social development. The research on information mining and propagation technologies in the microblog networks has a positive significance for microblog network marketing, public opinion monitoring and controlling untrue microblog opinion.Sina weibo data has been collected from Fudan University, Shanghai University and Shanghai International Studies University for research objects in this paper. The acquisition methods of user, tweets and concerned relations data among users in the microblog networks are introduced. An association algorithm for user authentication in the local area microblog network is proposed. An improvement average degree value 6)nn algorithm of the nearest neighbor is also proposed to determine the degree-degree correlation, which uses the median to compute the average in-degree of nodes. The microblog networks topology measurement and user behavior measurement are conducted from the perspective of empirical analysis, and then the microblog user relationship network with complex network characteristics shows a small-world effect and scale-free property.For the perspective of information dissemination, the concept of propagation matrix is defined and a new Weibo Rank algorithm for key node mining in the microblog networks is put forward considering parameters such as the user direct appeal, the scope and average information load. After comparison and evaluation on Page Rank, betweeness centrality, closeness centrality, Out-Degree, Weibo Rank five key node mining algorithms performance based on tweets propagation model, the results show key nodes obtained from the Weibo Rank algorithm have a wider transmission range and better influence.The ISMR(Ignorants-Spreaders-Multiple Retweeters-Rejecters) model is constructed with more fitting the real situation of microblog network information dissemination and the propagation mechanism of the model is introduced in detail. The propagation thresholds of ISMR model in homogeneous and inhomogeneous network are mathematically derived and the results show that no matter how small transmission rates are, the information can be spread out. With the ISMR model in the BA scale-free network and real microblog user relationship networks, four kinds of node density of ignorants, spreaders, multiple retweeters and rejecters changing with time are analyzed by simulation. A new WR immunization strategy is proposed and the influences of the immunity seed node density g and the delay time on the information spreading are studied by simulation experiments. Compared with random immunization and target immunization, simulation results show that WR immunization can more effectively suppress the spread of false information.The two main factors affecting the user forwarding behavior: user features and tweets features are discussed. Retweeting behavior is predicted using the SVM method and on this basis, the retweeting scale of the information is better predicted considering the attention relationship between users. Visual analysis on the microblog information dissemination can be more intuitive observation tweets propagation path. There are "dandelion", "double star" and "multi-pole" three kinds of information dissemination pattern. Discrimination categories method of information propagation model is put forward based on the relationship between the Weibo Rank values of each node.