The Design And Implementation Of News Recommendation System Based On Item Feature And Learning To Rank

With the gradual popularization of the Internet,online reading of news has become an important part of our daily lives.However,in the face of massive information on the Internet,people often become at a loss.As a means to effectively deal with the problem of "information overload",the recommendation system has been widely studied in industry and academia.This paper has studied the recommended methods in the field of news.The main research work of this paper is following:(1)Analyze the existing popularity prediction algorithm,and propose an improved news popularity prediction method by mining the characteristics of news.After using various clustering algorithms to cluster news,we will predict based on the time series of the same category of news.And on this basis,we improve the non-personalized recommendation algorithm;(2)According to the user's read behavior,we use the recurrent neural network to obtain the user's interest feature.And after the user data is acquired,we update the feature representation of the article for a specific user while updating the user's feature.Through the user feature and the article feature,we can generate a list of recommendations;(3)Based on the popularity prediction,we propose a recommendation method for improved collaborative filtering.Combined with(2)and multiple recommendation algorithms,we use learning to rank to fuse recommendation result for the top-N recommendation problem.Then we propose a hybrid recommendation algorithm based on learning to rank.We conducted experimental tests on public dataset and the result shows that the algorithm has better recommendation performance.(4)Based on the above proposed algorithm,we design and implement a news recommendation system based on item characteristics and learning to rank,which considering user habits and effectively responding to the user's cold start problem,enhancing user experience and improving recommendation accuracy.