A Real-World Library Dataset

This dataset contains real-world anonymized library transaction data collected from the integrated database of a university library. It covers book loan transactions recorded between 2018 and 2025 and comprises 52,394 borrowing transactions involving 14,148 library members and 16,217 library collections. It was compiled to support research on recommender systems, feature selection, machine learning, data mining, and library analytics. The dataset preserves the characteristics of operational library data, including interaction sparsity, data imbalance, and heterogeneous user-item attributes, making it suitable for evaluating recommendation algorithms under real-world conditions. All personally identifiable information has been removed or anonymized to protect user privacy while preserving the dataset's analytical value.