Research On University Libraries Personalized Book Recommendation Algorithms Based On OPAC

University library is regarded as the centre of the university literature. However, in this sea of knowledge, on one hand, the readers are difficult to find the literatures they are interested in, on the other hand, many lesser known literatures because of its large quantity cannot be accessed easily by users. Online public access catalog (OPAC) which is used to retrieve literature by users consists of a great amount of readers’information and retrieval information. This information can fully contribute to reveal the reader demand and the using situation the library resources. Based on the combination of the information and the recommendation algorithm, the staff can recommend the information that the readers require. By this way, the library can satisfy the customers better.This paper enumerates and compares several latest recommendation algorithms and analyzes the characteristics of the readers of University Library and the written resources. Based on the analysis, the author puts forward to use the "Chinese Library Classification" to solve the non-precision and cold boot of the key words and data sparseness and cold boot by setting up user-item scaling evaluation matrix in the collaborative filtering recommendation. Finally the author combines the two algorithms and put forward the book recommendation strategy which combine with the content-based recommendation algorithm and the collaborative filtering recommendation algorithm.Finally, the empirical research is based on the readers’ related information in the library’s OPAC database, and the experimental results show that the "Chinese Library Classification" improves the two recommendation algorithms and the book recommendation quality has markedly improved. The improved algorithm can effectively solve the synonym and antonym by use of the Content-based Recommendation algorithm. It also solves the problem of sparseness and initial evaluation problem in user-item evaluation existing in the collaborative filtering Recommendation algorithm.