Design And Implementation Of Personalized News Recommender System Based On Users' Interests And Preferences And Social Relations

With the advent of the Internet web 2.0,how to read interesting news has become an important part of the lives of many netizens.Then in the mass of Internet news,people are often forced to read the news pushed to us by editors.Therefore,recommendation system emerges as The Times require,and as an effective means to solve this problem,it has been widely studied in academia and effectively applied in industry.This paper studies the existing problems in the field of news recommendation.The main research work is as follows:(1)A personalized news recommendation method based on users'implicit social relationship is proposed.In the explicit social relationship of users,similar user groups are less,while the implicit relationship of users can mine more similar user groups.Based on this phenomenon,we constructed users' implicit social relationships from the news data set,and then modeled users' implicit influence probability based on poisson matrix,so as to generate users' news recommendation list.(2)In order to solve the problem that the features of users' interests and preferences are not constructed accurately,a personalized news method based on users' search records and interests is proposed.By constructing the exogenous user interest preference from the user's search record,the proposed fusion of the two preferences is used to generate the final recommendation list.Experimental tests on real news data sets show that this method has a good recommendation effect.(3)Aiming at the problem of cold start and data sparsity in news recommendation,a personalized news recommendation model based on deep network is proposed.Through deep network coding of users and news,a more mining and more complex relationship is built,which is verified on the real data set.Compared with the comparison algorithm,the recall rate is significantly improved.(4)Based on the above three algorithms,a personalized news recommendation system based on users' interests and preferences and users' social relationships is designed and implemented.The system includes three function modules:user function module,news display recommendation module and background management module.In addition,non-personalized functions are added to solve the problem of cold startup,and user experience is enhanced.