Research And Implementation On Information Forwarding Prediction Technology In Social Network

In social network,information forwarding refers to the process of information dissemination in which a certain information in social network is forwarded by different users.With the development of social networks in people's daily life,online social networks such as microblog have gradually become the basic platform for information dissemination and diffusion.This makes some false information have communication channels,and brings threat to the normal order of social network.Therefore,it is of great practical significance to predict the information forwarding behavior in social networks.In the current social network information forwarding prediction methods,the classical information propagation model is not suitable for user forwarding prediction,and the artificial intelligence classification algorithm does not consider the user network relationship and node attribute characteristics at the same time.To solve these problems,this paper studies and experiments the common attribute features of nodes in social networks,and proves the influence of topic features on information forwarding prediction.On this basis,a social network information forwarding prediction method based on weighted heterogeneous information network is proposed.Firstly,the topic distribution feature is introduced as the node weight to construct the weighted heterogeneous information network,and then the weighted heterogeneous information network is generated for low dimensional vector representation and forwarding prediction.In addition,considering that metapath2vec,a classical low dimensional vector representation generation algorithm for heterogeneous networks,is limited by meta paths in the process of random walk,and the network coverage is not enough,this paper optimizes and merges the results of random walk under multiple meta paths.Through a series of experiments on Sina Weibo data set,compared with the prediction results without considering the node characteristics and using metapath2vec algorithm,the accuracy of this method is improved by at least 6%,which proves its effectiveness.At the same time,this paper designs and implements a social network information forwarding prediction system,which includes four modules:offline database,data processing module,low dimensional vector representation module and information forwarding prediction module.This paper describes the main functions and processes of each module in detail,and tests the function and performance of the social network information forwarding prediction system.The test results show that the system can effectively realize the prediction of social network information forwarding behavior.