Design And Implementation Of A Microblog Water Army Identification System Based On Graph Neural Network

With the rapid development of mobile Internet and social network,social media has become an important platform for people’s daily entertainment,communication and sharing,and information acquisition.Sina Weibo,as the largest social media in China,affects all aspects of people’s life.At the same time,the huge flow of Weibo also gave birth to the black industry of microblog water army.Driven by interests,they spread spam or rumors,disrupt the order of network communication,guide public opinion,and bring adverse effects to individuals and collectives.How to effectively and quickly identify the Weibo water army is of great significance to purify the network environment and maintain the network order.Traditional research on network water army detection mainly focuses on content,user,environment and comprehensive characteristics.Due to the normalization of water army behavior,it is difficult for traditional methods to effectively and comprehensively identify them.In this article,a method of Weibo water army recognition based on graph neural network is proposed by using the social network structure which is difficult to forge.The specific research work is as follows:Firstly,Weibo users are regarded as the nodes of the graph,and the attention relationships between users are regarded as the edges of the graph.The data is obtained by web crawler,and the information of Weibo users and the association information between users are extracted to construct the social network graph of Weibo users.Next,the Weibo users are manually annotated and divided into five fine-grained categories: non water army,slightly suspected water army,suspected water army,severely suspected water army and confirmed water army,and the original data set is obtained.Then,in the process of data preprocessing,a new piecewise linear normalization method is designed to solve the problem of large data span of nodes,which makes the processed data more evenly mapped between [0,1],weakens the adverse impact on the model caused by some large variables.Finally,by studying the graph neural network algorithm,the identification model of Weibo water army is built.The graph is input into the graph attention network,and the feature vectors of nodes are updated through the stacked graph attention layer to obtain the node classification results of the graph,so as to achieve the purpose of recognizing water army.In order to verify the effectiveness of the model,the comparative experiments of graph attention network,logistic regression,naive bayesian,support vector machine and graph convolution neural network are designed.Four evaluation indexes are used to evaluate the effectiveness of the model.The experimental results show that the model has achieved good classification effect and strong generalization ability when the scale of labeled training set is small.Combined with this classification model,a Weibo water army recognition system based on Django framework is designed and implemented,and the recognition effect of the system is tested to verify the feasibility of the system.This system can easily and efficiently identify the water army,which is helpful to facilitate the identification or research of sina Weibo,and has a good application prospect.