It is worth studying and exploring the research on the personalized recommendation technique that is used in news reader application to help users fast access interested news with its rapid and accurate features. We design and implement a news recommendation system based on user collaborative filtering recommendation technology. By collecting user data, calculating the reading time factor to correct user preference value, incorporating the influence of the news heat and punishing user similarity value by the heat, and then conducting Top-N ranking for user’s unread news based on similar neighbor sets, the news of interest is pushed to the users. The test results of the news recommendation system  show that it can provide real-time updates for the user interest model accurately and each function has achieved prospective design aim.