Personalized recommendation by using fused user preference to construct smart library

Abstract

In online digital library, it makes people be convenient to visit latest books or magazines via digitization. Online digital library becomes the platform for people to search and browse latest books or magazines. From digital library to smart library, the key is how to recommend interesting items from personalized preference according to visiting history records, comments, and discussions. This paper introduces user preference into digital library recommendation system. Generally, if two users have similar preference, they also like the books or magazines from the same or related topics. First, the topics is estimated by a Latent Dirichlet Allocation (LDA) model; second, the topics with high user preference score is recommended to the user. Thus, the user can receive the item list which are close to the topics which he or she is interested in. The experimental results on a university digital library show the effectiveness of the user preference based recommendation system for smart library.