Research on the Performance of Collaborative Filtering Algorithms in Library Book Recommendation Systems: Optimization of the Spark ALS Model

Since the rapid development of information technology, the amount of book information stored in university libraries has been increasing. Traditional information retrieval techniques can no longer meet the needs of teachers and students to quickly search and filter information of interest from a massive collection of books. Therefore, in order to better serve university teachers and students, establishing a collaborative filtering recommendation system for university libraries can help push book information more accurately. This article introduces the development process of the Spark model, elaborates on the recommendation algorithm based on the Spark ALS model and its optimization, and studies the collaborative filtering recommendation algorithm based on the ALS model. The algorithm is implemented on a distributed platform, compared to previous single node implementations. The experimental results show that the optimization algorithm based on the Spark ALS model has greatly improved its computational speed, and reduces the loss of hidden factor item attribute information by integrating item similarity in the loss function. At the same time, an interest forgetting function is also introduced in the prediction score obtained from the optimal model. Through experimental comparison, the results show that the optimization algorithm proposed in this paper effectively improves the accuracy of the recommendation system. At the same time, a questionnaire survey was conducted on 250 students who used the system. The survey results showed that the students who participated in the internal testing had a good overall satisfaction with the system, with scores above 3.9 out of 5. The experiment showed that the improvement of the system was significant, and the actual application effect was good.