The Research On Network Information Propagation Model And Prediction Algorithm

In social networks,information propagation refers to the propagation process of information,such as opinions,rumors,comments,and user behavior,which are carried out by social networks.Social network plays a more and more important role in information propagation nowadays,and the research on information propagation of social network is of practical significance.The target of this research is to understand the inherent law of information propagation,and to timely warn of emergencies and sensitive information;to analyze group behavior and preferences of users,to predict the trend of user behavior.Therefore,the research on information propagation in microblogging network becomes more and more attention.Based on the methods of complex network theory,propagation dynamics,social psychology and machine learning,this paper studies the propagation and evolution of social opinions,the propagation of rumors,the prediction of propagation coverage and depth of microblog.The main work and innovation of this paper are as follows:To investigate the problem of Co-evolution of opinions and networks,a bounded confidence consensus evolution model in dynamic adaptive network is proposed.The evolution of opinion is affected by the networks,and in turn,the evolution of opinion also leads to dynamic changes of the networks.In order to study the evolutional rules of opinion in dynamic adaptive network,four statistical indicators,which includes the average number of opinion clusters,the probability of consensus,the proportion of the largest cluster,and the average number of running steps,are proposed.The experimental results show that the statistical indicators of the dynamic network model is better than that of the static network model,and the proposed model is more in line with the actual situation of the society.A rumor spreading model with forgetting mechanism considered is proposed in order to study the influence of social psychology on rumor spreading.The experiment results show that,the exponential forgetting rate can reflect the actual situation better: the forgetting rate has significant effect on the density of infected and removed,the greater the init ial forgetting rate or forgetting speed is,the weaker the rumor spreads;compared with the case of constant forgetting rate,the exponential function forgetting rate is more in line with the actual situation of rumor spreading.The correctness of theoretical analysis is verified by simulation experiments and the control strategy of rumor is also proposed.The research results are helpful to understand the behavior of rumor spreading and to provide useful reference for the spreading process and the prediction of network public opinion.In the research of forwarding prediction of microblog,whether a piece of microblog will be forwarded and what is its forwarding probability are studied most,the quantitative forwarding behavior is studied less.In order to solve this problem,an improved random forest algorithm is proposed to predict the propagation coverage of microblog.The extracted features are divided into three categories: user features,microblog features and social features.Random forests,decision trees,K-nearest neighbors and improved random forest algorithms are used to train the models.The experimental results show that when different number of data are used to test the improved random forest algorithm,the prediction accuracy fluctuates very little,the improved algorithm has high accuracy and good generalization ability.An improved random forest algorithm is proposed to predict the propagation depth of microblog.The extracted features are divided into three categories: user features,microblo g features and social features.Random forests,decision trees,K-nearest neighbors and improved random forest algorithm are used to train the models.The prediction accuracy of different algorithms is analyzed and the influence of different features on t he prediction results is studied.The experimental results show that social features have a greater impact on the prediction accuracy of propagation depth,and the improved random forest algorithm has higher prediction accuracy,therefore,that it has a high reference value.