Research On Prediction Of Public Opinion Influence Based On Social Network Aggregation Groups

In a free and open social network,the social public opinion behind the uproar is particularly important to understand the current social people's value orientation and emotional attitude,and has gradually become a weathervane for the healthy development of society.Therefore,the research on the influence prediction of network public opinion has become an important research direction at present.However,the distribution of user groups in the network public opinion has the characteristics of aggregation,the users within the group are very active,the information interaction between users is more frequent and timely,and the information flow is large.Therefore,how to establish efficient data processing methods,eliminate the false and retain the true of the complex and diverse public opinion data,obtain the influence of the current public opinion by analyzing the characteristics of the public opinion data and predict the development trend of the public opinion in the future has become the difficulty of the current research on the influence prediction of social public opinion.Therefore,according to the structural characteristics of social networks and the development law of public opinion,this paper proposes a prediction method of public opinion influence based on social network aggregative groups.Firstly,aiming at the problems of complex social network structure,large amount of public opinion data,time-consuming and inefficient calculation and analysis in public opinion events,this paper proposes a clustering algorithm OGC based on clustered user groups.Through statistical analysis of the relationship between the aggregation characteristics of users and user influence in social networks,simplify the network structure,narrow the research scope of public opinion influence prediction,and highlight the core group of public opinion events.Then,in order to improve the generality and flexibility of the algorithm,a parameter clustering granularity is further proposed to control the cluster size and connectivity between clusters.Secondly,on the basis of studying the clustering algorithm of clustered user groups,an adaptive user mining algorithm ASMO is proposed by analyzing the diffusion law of public opinion topics and the distribution characteristics of network node degrees.The algorithm takes the clustering results of OGC as the input,and then mines the most influential seed nodes based on OGC to extract the core users who dominate the development of public opinion.An adaptive parameter to control the number of seed nodes is proposed,which not only reduces the problem space of public opinion prediction,but also improves the ability of users to explain the influence of public opinion.Finally,the information propagation law of seed node set is introduced into the prediction of public opinion influence.According to the time series characteristics of public opinion data,a weighted difference method is proposed to stabilize the data.ASMO algorithm combined with time series prediction model ARMA and Prophet are used to fit and predict the public opinion influence of seed node set,and experimental comparison and verification are carried out.The experimental results show that based on ASMO algorithm's processing of social public opinion data,the prediction time of ARMA model is reduced by 9.53%;The prediction time of Prophet model is reduced by 13.03%.