Design Of News Recommendation System Based On User Browsing Mode

The Internet industry is growing rapidly nowadays,in the current age of big data,huge amounts of data makes people difficult to select the right information,lost in the ocean of information.Before the advent of recommender systems,people use search engines to find their information needs,however,in certain scenarios,the user fails to properly find the words they need,making search engine effects greatly reduced.Recommended system is an effective solution to the users’ information overload problem without explicitly demand,it has become a hot spot in many areas.Based on the context of big data,personalized recommendation system gets the user’s behavior,personal preferences by analyzing large amount of user data,provides timely and accurate recommendation results to users.Personalized recommendation system can intelligently provide users’ most interested contents to them,enable users to find what they need from the information.Personalized recommendation is also important for Internet news,news website like Today’s headlines site and SINA news post news from various trades and industries every day,along with the increasing number of news,users find it difficult to get the right news they are interested in.News recommendation system can detect users’ potential interest and obtain recommendation results base on users’ browse behavior and personal information,it can save users’ time and increase their satisfaction,also reduce the level of news resources waste.Current frequently-used recommended method largely based on explicit feedback data.However,different from the explicit feedback data,implicit feedback data is more accessible and universal,recommendation systems based on implicit feedback data has broader adaptability,this article will use implicit feedback data to design a recommendation system.This thesis studies the recommended model and algorithm,mixes different types of recommendation algorithms,designs a news recommendation system.Based on users’ browsing frequency,we divide the users into active and inactive users,we use collaborative filtering algorithm based on users and recommendation algorithm based on content for active users and collaborative filtering algorithm based on items and recommendation algorithm based on content for inactive users.After completing the design effort of recommendation system,we use evaluation to evaluate the recommendation system performance.