An Accurate Recommendation Method of University Library Digital Resources Based on User Interest

With the advent of the digital age, university libraries are faced with the challenge of how to efficiently manage and recommend huge digital resources to meet the diversified needs of users. Traditional recommendation system (RS) often ignores the evolution of user interests over time and the importance of personalized recommendation. This research aims to solve the above problems through the accurate recommendation algorithm of university library digital resources that integrates the evolution of users’ interests. This paper first discusses the background of the dynamic change in user interest and the urgent need for an efficient and accurate RS. The motivation is to better understand and predict user interests through a new algorithmic framework to improve the accuracy and user satisfaction of the RS. We propose a RS framework that combines the user interest evolution model and attention mechanism based on Transformer, and design a hybrid RS by combining content recommendation and collaborative filtering technologies. The system is designed to capture subtle changes in user interests and solve cold start problems, thereby providing more personalized and accurate recommendations. In the experimental part, we use the real university library digital resource dataset to verify. Experimental results show that compared with traditional RSs, our approach has significantly improved on several evaluation indicators such as accuracy, recall rate and F1 score, while also performing well in terms of diversity and novelty. This proves that our approach can effectively improve user satisfaction and interactive experience.