Research Of The Key Technology In User Mining And Message Propagation Prediction For Micro-blog Public Opinion

With the rapid development of social networks, Micro-blog has become an indispensable part of people's life. Compared to the traditional media, messages related to public opinion event in the Micro-blog spread faster. Micro-blog brings convenience for people to obtain information,but also provides a platform for criminals at the same time. If there are criminals using social platform to spread fake information, it will bring great inconvenience to people's life. Therefore it's important to monitor the spread process of the Micro-blog events, analyze key users who cause massive retweets and predict the scale of messages propagation in the future. It has a very important significance to control public opinion events to the negative aspect.In the propagation of public opinion events, the users have more attention in the public opinion events which are closely related to users in the regional level and cause user interest.However, the existing key user mining algorithms do not fully consider the regional characteristics of public opinion events. Based on this problem,this paper proposes a key user mining algorithm for Micro-blog public opinion - KURank algorithm, which is improved by PageRank algorithm based on the regional particularity and special description style of public opinion events. Meanwhile this paper adopts the concept of super network, which describes Micro-blog messages from four aspects of social sub-network, information dissemination sub-network, region sub-network and describing words sub-network.Experiments show that,KURank algorithm can effectively identify the key users in the Micro-blog public opinion event who promote the dissemination of public opinion events.In the process of Micro-blog information dissemination, large-scale propagation of events is usually caused by key users. In the research of message propagation prediction, the effect of key users should be an important consideration. Therefore this paper proposes the message propagation prediction model -- KU regression model based on key users. The model uses KURank algorithm to calculate the key users in each time window, constructs the polynomial regression model of Micro-blog trend prediction. Which uses the independent variable is ratio of new key users to all users and the dependent variable is growth rate of Micro-blog scale in each time window. Experiments show that the KU regression model can effectively solve the problem of public opinion event messages dissemination and the prediction results are in good agreement with the actual data curve.In summary,this paper emphatically considers particular geographic area and specific describing features of public opinion events. Based on the key user and polynomial regression function, key users mining model and message propagation prediction model are proposed.The feasibility and effectiveness of the KURank algorithm and the KU regression model are verified by experiments.