Design And Implementation Of News Recommendation System Based On Collaborative Filtering

With the continuous development of information technology,the information on the network is exploding,and the demand for information resources,especially news resources,is constantly increasing.Unlike traditional news,online news in the new media era breaks the limitations of traditional news timeliness and extensiveness and covers all aspects of life including economy,military,sports,and life.But each reader has a topic of his own interest.For each reader,those news topics that are not of interest are noise information,and the news recommendation system aims to satisfy their personalized interests by finding a group of news for readers.Solve the problem of information explosion.A very important issue with online news platforms is that the number of articles may overwhelm users.In order to alleviate the impact of information overload,it is important to accurately characterize and understand the users interests and tastes and to present personalized recommendations.This paper designs and implements a news recommendation system based on collaborative filtering.There are three major innovations in this paper:1.The article uses a collaborative filtering approach to use the Recurrent Neural Network(RNN)for news recommendations.However,in the face of the problem that the RNN network structure in the collaborative filtering is difficult to recommend in parallel,the slice neural network(SRNN)is creatively applied to the collaborative filtering recommendation system.Improve the training and operation eff-iciency of the system.2.Faced with the problem of cold start and data sparsity in collaborative filtering,this paper uses knowledge base(knowledge map)as auxiliary information of recommendation system.It not only solves the problem of cold start of the system and data sparsity,but also improves the accuracy and interpretability of the recommendation system.3.In view of the dynamic user preferences in the recommendation system,in order to increase the relevance of each item or feature level between the user s history and future interests,this paper uses the technology of enhanced memory(the external memory KV-MemNN is introduced in the model),Bnhance the recommended performance of the model by more clearly finding and updating the user s history.This paper first expounds the background and source of the project,and introduces the domestic and international research status of news recommendation.Later,relevant knowledge of news gathering,data processing,knowledge map,neural network and so on were introduced in the relevant technology research stage.In turn,a brief overview of system feasibility and functionality as well as non-functional requirements is provided.In the stage of system design and implementation,through the combination of requirements analysis,each function point is described in the module,and the internal logic of each function point is clearly described in combination with the class diagram and other diagrams.At the end of the paper,the design and implementation of the system are summarized,the direction of improvement is found,and the future development direction is prospected.