Intelligent book recommendation system for libraries supported by gray wolf optimization algorithm

Abstract

This paper presents a library book recommendation system designed to improve effectiveness, utilizing the GWO algorithm. The system architecture consists of three distinct layers: the foundational data layer, the data processing layer, and the intelligent service. The improved CGWO-KM algorithm is used to cluster project attributes, and the search and update mechanism of the gray wolf population is applied to find better initial clustering centers. Missing rating data is then filled in, and user similarity is calculated. A harmonized weighting factor is used to eliminate the correlation between ratings from different users. The weighted rating mechanism comprehensively considers both user ratings and the influence of neighboring users. The improved Pearson correlation coefficient combines the weighting factors with user similarity to obtain the final recommendation score, completing the intelligent book recommendation process for the library’s books. The results show that at the 5th month time snapshot, the predicted data (6) for the library’s historical borrowing dataset closely matches the actual value (3.7). The method proposed in this paper demonstrates an IGD mean close to the true optimal solution across various library datasets for literature, science popularization, history, art, and novels, with values of 0.0012, 0.0023, 0.0014, 0.0021, and 0.0020, respectively. The optimal non-dominated solutions in the three-dimensional space for resource utilization, recommendation diversity, and user engagement are close to the ideal value of 1. Moreover, the book recommendation system has a short processing time, and the recommendation accuracy ranges from 0.882 to 0.993, providing personalized, high-quality book recommendation services for readers.