Research On WeChat User Friend Recommendation Based On Link Prediction

For the friends recommendation process in the social networks,we generally use algorithms and model-building methods to calculate and to predict the hidden links between the social network users,then recommend the potential acquaintances or friends they may know or may be interested in by ranking the link level in descending order.The existing researches on friends recommendation in social networks provide us a variety of ideas.The most widely studied method is based on the users' interests preference,others including the method based on the users' social relationships,the method based on users'social influence,the method based on friends trust,the method based on users behaviors,the method based on geographic location,etc.Link prediction is a method of predicting the link possibility between two nodes in a complex network environment.Most of the existing researches consider the link prediction method with the traditional social networks such as blog and MicroBlog,but few focus it on the new decentralized social network like WeChat.Therefore,this article adopts this new perspective,conducting the WeChat friend recommendation by using the relationship between link prediction and the WeChat users nodes.This article analyzes the WeChat's systematic framework,structural design,and WeChat users' hierarchical relationship,and clusters the users' information and tags carried by WeChat users' personal homepage,moments and friends grouping status.Then we introduce the link prediction idea,that is,regarding the WeChat users as nodes and their relationships as edges.We refer two features of link prediction,the nodes similarity and the nodes influence,and use the TOPSIS model to achieve the matric operations and to conduct the WeChat users friend recommendation.Finally,this paper uses open source code to grab a small amount of data on WeChat Moments whose users have opened the Root permission from two WeChat groups.By analyzing the experimental results,we can find that the link prediction algorithm based on nodes influence has higher accuracy than other algorithms,thus verifying the feasibility of WeChat users friend recommendation method based on link prediction.