Design And Implementation Of Friends Recommendation System Based On Sina Weibo

With the rapid development of Internet and mobile communication technology,more and more people use Sina Weibo and other social platforms to make friends and share something.Hundreds of millions of users communicate online to produce a huge amount of data,so that the phenomenon of "information overload",that making people spend more time looking for friends than to communicate with friends.To solve this problem,this thesis designs and implements a friend recommendation system,which can recommend friends for users.This thesis obtained the personal information and Weibo information of the second-degree friends of the target user through the way of web crawler,and analyzed the collected data based on the user interest similarity,the user geographical similarity and the user influence to comprehensively recommend friends for the target user.Firstly,this thesis introduces the research background and significance of the subject,and analyzes the research status of the subject at home and abroad.Secondly,through analyzing the user requirements and functional requirements of the recommendation system,the thesis finishes outline design of the system,divides the function modules of the system and designs the database of the system.Thirdly,it makes detailed design and implementation of every function module in the system.Among them,Weibo data acquisition module implementates a Sina Weibo crawler based on the concern relationship between users.Through the breadth first search friends,the crawler obtains the second-degree friends of the target user,through the analysis of the web page to obtain friends's personal information and Weibo information,completes the data persistence,at the same time solves the problem that Weibo open API access to a variety of restrictions on data.The friend recommendation module extracts the text feature words by using the Ansj Chinese word segmentation and the TF-IDF algorithm to Weibo history content,and the classification of feature words to obtain user interest vector by using Naive Bayesian classification algorithm,and calculates the user interest similarity by cosine distance.Then,the distance between users is calculated by the user's location information and the user's check-in data,and the distance is converted into geographical similarity,and the geographical similarity is normalized by the normal distribution function.After that,the user influence is measured by the number of fans of the user,the number of Weibo,and the amount of Weibo forwarding,the amount of comments,and the amount of like.Finally,by assigning different weights of the three factors and combined with the user's educational background and work experience information to generate a recommended list,then recommend friends of target user by the Top-N method.The experimental results show that the multi-factor friend recommendation are more accurate than the single factor.