Analysis And Prediction Model Of Hot Topic Influence In Social Network

With the rapid development of the Internet era,online social networks have become an important channel and carrier for people to spread hot topics,which has greatly improved people's life services and quality.At the same time,due to the complexity and multi-dimensionality characteristics of topic participation users' relationship and behavior in online social networks,it is difficult to effectively monitor and control the information participation process.How to effectively mine the key element information in the hot topic participation process in the whole life cycle,and how to effectively predict the importance of hot topic influence in the future process of hot topic participation,are an extremely practical research direction.This thesis combines the theoretical basis of ternary association graph model and heterogeneous node propagation network to analyze and predict the influence of hot topics in social networks.The main work and contributions of this thesis are as follows:1.In terms of topic influence analysis,focusing on the path,user and domain of social topic participation in the whole life cycle,this thesis proposes an analysis model of multi-domain and multi-stage social topic influence based on ternary association graph.Firstly,a path-user binary association graph model is extracted and a user-domain division model is constructed.Then,a path-user-domain ternary association graph model is constructed,and the potential relationship among the three key elements is described.Finally,the influence analyzing algorithm based on ternary association graph and cross-iteration scoring mechanism is proposed,and the information of each key element is mined.As the same time,the propagation situation of each key element in the whole life cycle is analyzed dynamically.2.In terms of topic influence prediction,combining heterogeneous node propagation network and exponential smoothing prediction model,this thesis proposes a prediction model of social topic influence based on double-weighted social network nodes.Firstly,the internal driving features and external driving features are extracted from the users' behavior data and users' relationship data,and the users' propagation willingness is quantified based on the internal and external driving features.Then,based on the extracted driving features,an extended social network based on double-weighted nodes was constructed,and the improved Page Rank algorithm model was used to iterate on this network to identify the users' propagation ability.Finally,combining users' propagation willingness and users' propagation ability to mine the key propagation users and the sequence of hot topic influence in each stage,and based on this sequence,a dynamic third-order exponential smoothing model is established to predict the importance of hot topic influence in the next moment.This thesis uses real Sina Weibo data to experiment with the above model.Experiments results show that the analysis and prediction model of hot topic influence proposed in this thesis can effectively mine and predict key element information,and significantly improve the accuracy and convenience of the analysis and prediction model.