Design And Implementation Of Personalized News Recommendation System Based On Knowledge Graph

With the application of personalized recommendation algorithm in the industrial sector,the Internet plus products are increasingly offering recommendation services.The importance and function of recommender system are more and more important.As a part of recommendation field,news recommendation has also become a research hotspot.A good personalized news recommendation system can bring new user traffic on the one hand,on the other hand,it can also consolidate the existing user base.Generally speaking,personalized news recommendation mainly includes two parts:generating recommendation candidate set and sorting candidate set.In this paper,knowledge map feature learning is applied to the recommendation system to design and implement personalized news recommendation system,and the candidate ranking model is studied.In this paper,the B/S architecture is used to build the system,the back end uses the Flask framework,the database uses Mysql,and the front and back end uses HTTP for data transmission.The main work is as follows:(1)This paper makes an empirical analysis and comparison of three recommendation models based on knowledge map feature learning,and selects the appropriate model to apply to the recommendation system.These three models correspond to the three methods of knowledge map feature learning applied to the recommendation system: Learning in sequence,alternating learning and joint learning.In the same experimental environment,using verifiable data sets,training and prediction of the recommended model are compared,using the recommended evaluation indicators AUC and F1 to evaluate the model,and finally the DKN model is selected;(2)Combined with the specific requirements and related technologies of the recommendation system,the overall architecture of the system is designed.In this paper,personalized news recommendation system is divided into news display analysis module,news recommendation module,data collection module,and the detailed design and implementation of each module are described.Finally,test the system and verify the effectiveness and availability of personalized news recommendation system.