Detection of Book and Borrower Communities Based on Book Borrowing Records in the Library Using Complex Network Analysis

Abstract

To improve user services, libraries need to have deep insight into users and the books they manage. This study aims to identify the community of users and borrowed books based on book borrowing records using complex network analysis. It starts with collecting book borrowing data, then converting it into a bipartite book-borrowing network using python programming and visualizing it with Gephi. Network analysis is performed to investigate network properties, and to use the BIMLPA method to find communities. The results of the investigation show that the structure of the book borrowing network is divided into two separate components. One main component of the network represents the optimal process of borrowing books, and the rest consists of many small components representing suboptimal borrowing. Community detection on the main component found 217 borrowing communities and 141 book communities. Most of the communities are small, where the book community is 2-4 members, while the borrowing community is in the form of individual borrowers. This research also produces the top 5 book communities, and borrower communities, the most popular books and the most active borrowers. The characteristics of the users and the books found can be used as a library reference for a more effective and efficient book development policy strategy, as well as book recommendations for more targeted users.