Abstrack -Book lending is the most important service in the library. So far, book borrowing data is often used as a statistical report, has not been analyzed further to find patterns/knowledge to deepen the insight of library managers. With the rapid growth of big data, social network analysis and community detection have been studied intensively by many researchers over the past few years. However, little research has been done on social network analysis and community detection of borrowing books at the library, and no one has even conducted a comparison analysis of community detection algorithms on book lending. In this paper, we propose an analysis of the library's book borrowing database using social network analysis and community detection methods. The purpose of this study is to find book clusters and borrower clusters by utilizing the best community detection method obtained. The research step begins with collecting data on borrowing books, constructing it into a bipartite graph model, projecting the bipartite graph into a book graph and a book borrowing graph. Then conduct experiments comparing several community detection algorithms for the two graphs, with evaluation metrics in the form of modularity, performance, coverage, density and entropy. The experimental results of Louvain's algorithm and Eva's algorithm have the best performance for book graphs and book borrowers. The application of community detection to the book graph obtained 16 clusters of books, while the book borrower graph obtained 21 clusters of book borrowers. The results of this clustering can be used as recommendations for library management in making library programs to increase the utility of books and increase user loyalty.