Book Recommendation Based on Book-Loan Logs

摘要

Information overload affects the efficiency of resource utilization. To tackle the problems, book recommendation is one of the solutions for university libraries which possess huge volumes of books and reading-intensive users. The users borrow books mainly for course-learning and academic-studying, which means the recommendation strategy should consider not only personalization but also the commonness driven by curricular necessity. This paper first studies the users behaviors on a large scale book-loan logs of a university library; then implements two recommendation algorithms based on the book-loan data, one of which is the classical item-based cooperation filtering recommendation algorithm, the other is a probability-based algorithm proposed in this paper. The average precision of the probability-based algorithm performs better in a random sampled testing set. The paper finally discusses the application cases of different algorithms in university libraries' routine work.