针对传统书目检索系统的相关度排序存在的读者体验差的问题,提出基于BM25模型与借阅预测模型的排序算法。该算法综合了BM25模型的相关度算法和借阅预测模型的推荐度算法,使排序结果受到相关度、入藏时间、借阅情况、在馆情况和馆藏地点权重等因素的影响。测试结果表明,综合排序算法使结果排序更加科学和合理,能够提高读者满意度和图书的借阅率。 In order to solve the problem of poor reader experience in the ranking of traditional bibliographic retrieval system, a ranking algorithm based on BM25 model and loan prediction model is proposed. The algorithm combines the relevancy algorithm of BM25 model and the recommendation degree of loan forecasting model, which makes the ranking result affected by factors such as relevancy, entry time, loan status, the situation in the library and the weight of the collection location. The test results show that the comprehensive ranking algorithm makes the ranking of the results more scientific and reasonable, which can improve readers’ satisfaction and the borrowing rate of books.