Design And Implementation Of News Recommendation System Based On Deep Learning

With the advent of the Internet era,major websites and mobile news apps have joined the ranks of information push and information dissemination.Due to the popularity of online news,the public is no longer limited to obtaining hot current affairs information from traditional media such as TV and radio.In terms of the promulgation of policies and regulations or the implementation of relevant regulations,the news system of online media will spread faster than offline media,and reach the public faster and more widely.In addition,the news system can better pay attention to and understand the public.The ability to monitor and analyze public opinion in real time,are of great significance to social stability.However,the massive amount of information generated daily on the Internet makes regular news websites unable to meet the diverse needs.Based on the user-centered concept,major news websites have introduced recommendation technologies.On the one hand,compared with traditional recommendation methods,recommendation technique based on deep learning has unique advantages in extracting the deep semantics of users and items.On the other hand,it can mine rich user behavior and individual demand by mapping data to high-dimensional space for fusion.In this context,the article studies the news recommendation system fused with deep learning algorithms to provide a good news recommendation service for the majority of users.Based on studying the related theory and technology of deep learning,this paper proposes a system recommendation model MIND_EbNR.This model uses the noise-reducing autoencoder with category information and the gated recurrent unit to model news and users respectively,and uses the inner product of the two to measure their correlation.In order to verify the effect of the model,experiments were carried out based on the public data of Microsoft named MIND,and evaluation indicators such as AUC and F1 Score were used to measure and evaluate the experimental results.This article completes the development of the news recommendation system in accordance with the basic process of software development,combined with the above-explored model.The system is developed with a separation of front and back ends.Data is crawled through Selenium tools.The front end uses the Vue framework and the back end isimplemented based on the Spring Boot and MyBatis framework.Finally,the function of system such as authorization authentication,user management,news management,news loading,and user preference expression are realized.In this system,users not only can accept personalized recommendations and hot news recommendations from the system,but also actively search for information they are interested in.