Research On Popularity Of Microblog Based On Retweet Sequence Of Timestamp And User

As one of the popular social networks,Weibo generates tens of thousands of news every day.The hot events spread explosively on Weibo.How to predict the future popularity of Weibo in the early days of Weibo propagation has become a very challenging question.The challenge is that it is difficult to measure various factors known to affect the popularity of the content,such as the quality of the content or its relevance to the user.Popularity prediction problems also have significant practical value in commercial and administrative applications,such as public opinion monitoring,online marketing,and the like.Traditional methods of popularity prediction are generally based on classification or regression models,point-random processes,and infectious disease models.Based on the classification or regression model,the feature selection is heuristic and the extraction process is very tedious,and the dynamic propagation process of Weibo cannot be described.Based on the point-random process model,it is generally impossible to use historical message supervision,and performance will be lacking.Mathematical methods are used to model based on infectious disease models,ignoring the network structure between users.Aiming at the above problems,this paper proposes a time-based and user-based forwarding sequence based on the research and learning of microblog's own propagation characteristics(Popularity Prediction model based on Retweet Sequence of timestamp and user,PPRS).The performance of the proposed model and the three benchmark methods are compared experimentally,and the validity of the PPRS model is proved.The main research work of this paper is as follows:First,this analyzes and summarizes the existing methods of popularization prediction,and analyzes the advantages and disadvantages of three prediction methods based on classification or regression model,point-based random process and infectious disease model.Second,in this paper,we use the characteristics of the cyclic neural network to model the forwarding sequence of Weibo.The forwarding time and forwarding user are represented as vectors in each time step.Through the cyclic neural network,the historical supervised information can be effectively utilized,and then the rate of each forwarding moment is learned through the middle layer,and then the trend change of the microblog in the early stage,that is,the trend acceleration,can be calculated.At the same time,based on various analyses,this paper finds that user activity has a great influence on the popularity of Weibo.Therefore,this article quantifies user activity and adds it to the model to improve the accuracy of the model.Last,this paper first proposes a prediction model of microblog popularity based on time retweet sequence,models time information in microblog retweet sequence,learns microblog trend acceleration,and combines with user activity and early microblog popularity.,predict the final popularity of Weibo.Then consider the influence of the user in the process of communication,integrate the user's interests into the model,and propose a time and user-based microblog popularity prediction model,which not only utilizes historical communication information,but also portrays Weibo well.The process of dissemination has a strong flexibility.It is also proved by comparison experiments that the model has better prediction performance on the popularity prediction problem.