Design of Book Recommendation System Based on Machine Learning in Smart Library

This study aims to construct a book recommendation system based on ML (Machine Learning) in a smart library. We focused on the performance of traditional algorithms and DL (Deep Learning) algorithms in book recommendation, and evaluated their applicability in a smart library environment through comparative analysis. DL technologies such as embedding layers, multi-layer perceptrons, and attention mechanisms were introduced in the experiment to improve the system's ability to learn user interests and book features. This article also conducted experiments on response time and percentile response time, and the research results showed that the average response time of the recommendation system is 120 milliseconds, which meets the requirements of real-time recommendation. The 95% percentile response time is 180 milliseconds, indicating that most users can receive recommendation results in a relatively short amount of time. The introduction of DL technology will help smart libraries better understand and meet the personalized reading needs of users, thereby promoting the development of library services towards a more intelligent and efficient direction.