Research And Implementation Of Naval Group Recognition Based On User Characteristics And Community Discovery

In recent years,with the rapid development of social platform and mobile Internet,some organizations or individuals employ organized water army teams on the platform to make profits through hype,malicious comments,dissemination of untrue speech and other acts,which seriously damages the order of cyberspace and affects social stability.Therefore,the identification of online water army groups not only helps to maintain the social network order,but also helps to analyze the behavior purpose and define the responsibility of different water army teams.In this paper,aiming at the problems of insufficient feature differentiation,low coverage rate of recognition model and inaccurate community division in the research of water army group recognition,the research on Water Army feature construction,water army recognition model and group division is carried out.The main work includes:First of all,in view of the current lack of distinguishing features between Navy and normal users,which leads to the low accuracy of recognition model,seven extended features are constructed by considering the differences between Navy and normal users in account information,daily behavior rules,published content and social relationship.Through the analysis of the characteristics of water army data and the comparison of the characteristics of different algorithms,the random forest algorithm with diversity optimization is selected as the water army seed recognition method.In order to further improve the coverage rate of the model,the SD value of water army suspicious degree is defined to measure the water army possibility of each node,and the SD value of each node in the network is propagated and quantified by trustrank algorithm to realize the mining of suspicious water army.Finally,the effectiveness of the extended features and the proposed model is verified by experiments.Secondly,aiming at the problem that the accuracy of the results is not high due to only relying on the feature attribute clustering of the Navy,this paper analyzes and quantifies the similarity between the Navy users,and combined with the social network relationship,uses the improved Louvain algorithm to divide the Navy network.The experimental results show that the model proposed in this paper has a good effect on the recognition of Navy groups.Finally,in order to verify and display the research content of the text,the paper designs and implements the network water army group identification and analysis system.The system mainly includes three functional modules:Water Army identification module,water army group identification and display module,and user management module.The system realizes the visual display of water army group identification,group division and group membership relationship of each water army node.