Research on Personalized News Recommendation System Based on Collaborative Filtering Algorithm
Hua Jiang · 2022
The news that users are interested in changes over time. The current personalized news recommendation system does not consider the timeliness of news, which affects the recommendation effect. A personalized news recommendation system is designed based on collaborative filtering algorithm. In the hardware part, FPGA is used as the core logic control chip, the crystal oscillator output is directly connected to the global clock pin of FPGA, and the high and low level of output data is determined by the data byte control pin (LB, UB). In the software part, the overall architecture of the system software is designed based on the user log, and the personalized news recommendation model is established based on the collaborative filtering algorithm. The time decay function is introduced into the model to obtain the interest model belonging to each user, and find the news similar to the user’s preferences and push it to the user. The test results show that the recommendation effect of the system is the best when the number of users’ nearest neighbors is 30. The MAE of personalized news recommendation system based on collaborative filtering algorithm is 0.7488, which is 0.0648 and 0.0795 lower than that based on association rules and knowledge map. Therefore, the design system has better recommendation effect.