Research And Application Of Social Network Users' Follow Relationship Prediction Based On Information Retweet

In recent years,with the popularity of the Internet and the continuous development of information technology,the number of users using social network platform to communicate and obtain information is increasing year by year.The relationship between users affects whether users can obtain information in time and the scale of obtaining information.Therefore,it is particularly important to study the relationship between users in social networks.In social networks,the more stable relationship between users is the follow relationship.In the process of layer by layer dissemination of microblog information,other users know the blogger by reweeting the microblog information,thus causing follow to the blogger.Other users can get the information they want by paying attention to the bloggers.This thesis studies Sina Weibo users,one of the most used social network platforms in China,and excavates potential follow users from reweet users.At present,there are few literatures on the relationship between microblog users,and they do not start from the perspective of users' information reweet.Based on this,this thesis proposes a prediction method of social network users' follow relationship based on information reweet.The main research content is divided into the following three aspects:1.Users' follow characteristics are mined.Considering the factors affecting the follow relationship from the perspective of bloggers and users,the effects of relationship strength and information reweet on the follow relationship are obtained.Among them,the relationship strength considers the number of users' follow,the number of users' @ and the number of bloggers' fans,and the information reweet is based on the reweet habits of users.2.Reweet characteristics of users are mined.Information reweet divides users' reweet habits into reweet behavior and reweet content,analyzes the influence of users' reweet habits on the follow relationship,and takes users' reweet behavior and reweet content as the characteristics of follow prediction.Among them,reweet behavior considers reweet behavior preference,reweet activity and reweet chain depth,and reweet content considers interest similarity and topic tag.3.The user's concern prediction is realized.In order to make the effect of attention prediction more accurate and stable,this thesis uses an ensemble learning algorithm combined with majority voting method to experiment.Compared with a single classification algorithm and the set comparative experiments,the integrated learning algorithm is more accurate and stable in predicting the potential follow of reweet users.