Research On Key Technologies Of Network Hot Topic Situation Awareness

With the increasing of Internet technology and the rapid popularization of mobile terminals,the proportion of online social networks in people's daily life is increasing.Research on situational awareness technologies for hot topics in social networks can grasp the rules of dissemination of topic information.On the one hand,it helps deepen the understanding of network structure and users' behavior.On the other hand,it plays a good role in guiding public opinion of the Internet.So researching hot topic situational awareness has important value on theoretical research.From the perspective of online social networks,this article combines the main characteristics of social network including diversification,large-scale data,dynamic evolution and non-linearity,aiming at the behavior of user groups involved in hot topics and the change of hot topics in macro perspectives.The main research work and contributions of this article are as follows:1.Aiming at the complicated participation behavior of participating users under hot topics in social networks,a prediction model on users' behavior was proposed.First of all,we selected the personal driving mechanism and the social driving mechanism and extracted the influence factors of the user participation behavior.Second,we used RBF(Radial Basis Function)neural network to build user forwarding behavior prediction model.Since the mapping relationship between user attributes and behavior is uncertain,cloud theory introduced in fuzzy mathematics is used to optimize it and a C-RBF neural network(C-RBFNN)prediction model is proposed,expressing the ambiguity and randomness of user's forwarding behavior,and also having a good ability to fit nonlinear relationships.2.For the trend of hot topics,the application of the exponential smoothing method can accurately and objectively predict the changes in the topic of Weibo,making reference and preparation for the next step of decision-making.Due to the complexity of topic data,the exponential smoothing method can effectively eliminate the influence of data fluctuations and fit the data trend.Secondly,based on the traditional exponential smoothing model,according to the dynamic change of the topic data,the smoothing coefficient is adjusted according to the actual trend of the topic data during the prediction process,so that the prediction model can track the trend of the data in the prediction and reach high adaptive effect.Finally,experimental verification was performed using real datasets derived from Tencent Weibo.Experiments show that the model and algorithm used in this thesis can not only combine the topological structure and user attributes in online social networks,but also analyze and understand the role of hot topics and propagation mechanisms from a macro perspective,and predict the dynamics of hot topics.The trend of changes can play an important theoretical guiding role in the guidance of online public opinion.At the same time,it also has important application value in topic information recommendation and online marketing.