Collaborative Book Recommendation Based on Readers' Borrowing Records

Abstract

Book recommendation is an important part and task for personalized services and educations provided by the academic libraries. Many libraries have the readers' borrowing records without the readers' rating information on books. And the collaborative filtering (CF) algorithms are not proper under this circumstance. To apply the CF algorithms in book recommendation, in this paper, we construct the ratings from the readers' borrowing records to enable the CF algorithms. And to evaluate the traditional CF algorithms, we show that linearly combining (blending) a set of CF algorithms increases the accuracy and outperforms any single CF algorithms. At last, we conduct the experiments based on the real world dataset and the results invalidate the efficiency of the blending methods.