Research on optimization of library book recommendation system based on the collaborative fusion of transformer architecture and adaptive extreme learning machine

With the rapid development of information technology, the demand of library book recommendation system is increasing day by day. Traditional recommendation algorithms often rely on simple collaborative filtering or content-based methods, resulting in insufficient recommendation accuracy. In this paper, an optimization model of library book recommendation system based on Transformer architecture and adaptive extreme learning machine is proposed. In the study, we analyzed the database of a large public library, with a sample size of 100,000 books and 50,000 users, covering users' borrowing history, book attributes and scoring data. Transformer model is used to extract latent features of books and users, and adaptive extreme learning machine is used to perform feature fusion and classification to improve accuracy of recommendation. We contrast the recommendation accuracy between traditional recommendation system and the recommendation accuracy based on our proposed method. Results show that the recommendation accuracy of the fusion model is improved by 15%, the recall rate is improved by 20%, and the F1-score reaches 0.85. These data show that the self-attention mechanism of Transformer architecture can effectively capture complex relationships between books, while the adaptive extreme learning machine improves the learning ability and generalization performance of model.