摘要:As the first incremental association mining algorithm, FUP can well solve the problem, but the algorithm also has the deficiencies to produce a large set of candidates and multiple iterating the database, leading to the algorithm’s low execution efficiency when dealing with some large transactions with fast updates, such as book circulation data. This study proposes an improved FUP algorithm that takes transaction identifier (TID) in the database to scan the database only once, making the computation significantly less than the FUP algorithm. Through detecting the circulation data of a university library, the experimental results show that compared with the standard FUP algorithm and SFUA algorithm, with the increase of borrowing and record transactions, the improved FUP algorithm has significantly improved the operation efficiency, which can help the library to do a good job in book recommendation scientifically.