Method Of Single Detection And Group Division Of Internet Hirelings On Weibo

Not only both the order and security of cyberspace are destroyed,but also social stability is threatened by the negative influence of Internet Hirelings.Now,the management and control of Internet Hirelings have become a research hotspot in many areas.It is of great significance for the study of Internet Hirelings to analyze the differences between Internet Hirelings and normal users and to show an individual identification model of Internet Hirelings.Besides,the study on the group division of Internet Hirelings will contribute to the exploration of their behavior patterns and help define their responsibilities.In this article,the user information on Weibo is taken as the research object.After analyzing the drawbacks of the researches,this article focuses on the feature and construction,the individual identification algorithm,and the group division of Internet Hirelings on Weibo.The main contents include:Firstly,according to the lacking coverage of Internet Hirelings' features,four new features were built without some irrelevant ones removed by the chi-square test algorithm: user activity,content legality of Weibo,reliability of user identity and rationality of user behavior.The social science definition of Internet Hirelings and the differences between them and normal users on Weibo are integrated by these new features.Besides,they are formed based on user's characteristics of personal information,Weibo content,behavior,and relationship combining different prominent features of Internet Hirelings on Weibo,such as zombie account,hijacking paid-poster account,and core account.The experimental results show the feature set constructed in this thesis can better express the difference between Internet Hirelings and normal users,and effectively improve the effect of the recognition model.Secondly,to solve the difficult labeling of Internet Hirelings,APDHW(Base on Affinity Propagation Method of Single Detection of Internet Hirelings on Weibo)is proposed using semi-supervised learning.First,the label set of Internet Hirelings is expanded by introducing the Radius threshold into Affinity Propagation clustering with a part of labeled data and massive unlabeled data.Then,the identification of the Internet Hirelings account is realized by combining the support vector machine.The experimental results show that the proposed individual recognition algorithm is more suitable for identifying Internet Hirelings on Weibo.Thirdly,a group division method based on the account information of Internet Hirelings is proposed given the deficiency in current research.First,based on the original characteristics of Internet Hirelings,three new characteristics are constructed combined with Internet Hirelings' different roles and divisions in information dissemination: the activity,the influence,and the importance.Then,to find a more suitable algorithm for group division of Internet Hirelings on Weibo,Affinity Propagation,K-means,Density-Based Spatial Clustering of Applications with Noise(DBSCAN)and Mean Shift are selected as candidates in the model's community division algorithm.The experimental results show that the feature set constructed in this thesis can express the different roles and divisions of Internet Hirelings groups more effectively,and the K-means algorithm is more suitable for the group division of Internet Hirelings on Weibo under the current data scale.