Research On Group Behavior Of Social Network Based On Network Relationship

In social networks,research on user group behavior and information dissemination mechanism has become a hot topic of current research.With the continuous progress of society and the rapid development of the Internet,online social networks have become an important way for information sharing and rapid dissemination.The online behavior of users in social networks and the research on the law of information dissemination are of great significance for public opinion management,network marketing,and discovery of key users.This thesis analyzes the behavior patterns of users from the two levels of explicit links and implicit links,and perceives the trend of information dissemination.From the level of explicit link,this thesis constructs the network topology through the friendship relationship in the social network,and predicts the user behavior based on the ICS-SVM method.From the level of implicit links,this thesis finds that the factors affecting the user participation in the topic discussion are not only related to friends around the user,but also related to non-friends whose users are closely related or have similar interests.Therefore,this thesis uses machine learning methods to construct implicit links,improve network topology,combine infectious disease models,and perceive information dissemination trends.The main research work is as follows:1.At the level of explicit link,the research on the prediction of user forwarding behavior in social networks is mainly focused on the attributes and characteristics of social networks.The problem of the influence of user friends and the regularity of user historical behavior is not fully considered.Take into account the problem,this thesis designed a user behavior prediction method.Firstly,according to the captured microblog data,the user own attributes and friend attributes in the social network are extracted.Secondly,the improved cuckoo search algorithm is used to select the optimal parameters to optimize the SVM,which can predict the user forwarding behavior more accurately.Finally,according to the characteristics of the user forwarding behavior changing with time,this paper uses the method of time slicing to sense the information dissemination trend.2.At the level of implicit link,the current researchs focus on the explicit friend topology,ignoring the influence of implicit friends on information dissemination.This thesis designs a dynamic model of information dissemination.Firstly,based on factors such as the closeness of interaction between non-friends,the classification algorithm is used to determine whether there are implicit links between non-friends,and the network topology is improved by explicit links and implicit links.Secondly,in the improved network topology,the individual driving mechanism and friend-driven mechanism are extracted,and the two information dynamics of individual influence and friend influence are analyzed.A multivariate linear regression model is proposed to measure social influence and analyze the trend of information dissemination.Finally,in the infectious disease model,considering the timeliness and uncertainty of information dissemination,the mean field theory is introduced,therefore,this thesis obtains a model of information dissemination based on social influence.This thesis uses real data of Tencent microblogging to verify the proposed model and method.Experiments verify that ICS-SVM method can predict user forwarding behavior more accurately,prove the role of implicit link in affecting information dissemination,explore the dynamic genesis of information dissemination,and perceive the trend of information dissemination.