Prediction Of Forwarding Behavior Based On Social Network Users

Social network has become an indispensable part of people's life.Through in-depth research on users' forwarding,information transmission mode on the Internet can be understood,which has great research significance in public opinion monitoring,advertising push and other aspects.The prediction of users' forwarding behavior is mainly based on model selection and feature mining.The former can be studied from factors graph model,single-class collaborative filtering and matrix decomposition,and the latter can be divided into network structure,information content,user behavior and mixed features.This thesis mainly studies users' forwarding behavior in Twitter social network.In reality,users' forwarding behavior is determined by a variety of factors,including users' forwarding habit,the topic heat of tweets and the influence of the authors of tweets.When studying the influence of historical tweets on forwarding behavior,the timeliness of historical tweets is not taken into account.The research on emotion lacks pertinence and there is no emotion dictionary in specific field.The key to predicting users' forwarding behavior lies in how to excavate the factors affecting users' forwarding behavior more comprehensively and combine them effectively to improve the accuracy of forwarding prediction.In view of the above problems,the following works are done in this thesis:(1)Current studies on influencing factors of Twitter users' forwarding behavior mainly focus on user behavior,influence and content of tweets,while the prediction of forwarding behavior focuses on static features.This paper analyzes users' forwarding behavior based on sociology,including dynamic characteristics such as the distance between users and the author of tweets and the concern relationship between users.Mutual information and left and right entropy are used to construct an emotion dictionary in the field of twitter retweeting to improve the accuracy of sentiment word classification.(2)When mining users' interests and hobbies through historical retweets,the timeliness of historical retweets is ignored,and users' interests are constantly changing.This paper presents a forward user behavior prediction model based on process of hawkes,forward user tweets strength comes down on the basis of strength and motivation,and use the hough model calculating distance between users and tweets author,user number of forward forward on the basis of tweets,forward solve historical records for the current forward tweets,users interested in issues such as the influence of dynamic attenuation.Experimental results show that compared with Ada Boost model and RBMHDRN model,the accuracy of the proposed model is improved by 2.29% and 1.55% respectively(3)Build a user analysis system based on the prediction of twitter users' forwarding behavior.This system can effectively analyze and study twitter users' interests and regional influence,and provide intelligent support for research work such as public opinion control,accurate recommendation and user influence evaluation.