A Novel FAHP Based Book Recommendation Method by Fusing Apriori Rule Mining

Abstract

Book recommendation is becoming increasingly significant library service, considering it improve access to relevant books by making personal suggestions based on previous examples of user's preference. Most existing approaches are either collaborative-filtering based, considering the data sparsity and cold-start problems, collaborative-filtering approaches suffer from many challenges. In this paper, we present a Fuzzy Analytical Hierarchy Process (FAHP) based method by fusing Apriori rule mining. Apparently, multiple factors (e.g., similar preference, professional background, education degree and book's publishing house etc.) may influence reader's borrowing decision. Therefore, we first adopt Apriori algorithm to develop association analysis for evaluating the relevance of books in terms of book-loan history. Second, FAHP takes the result of association between books and other subjective/objective factors into account and makes final recommendation according to an overall ranking result. A thorough experimental comparison, based on real-world data, illustrates advantage of our scheme over collaborative filtering approaches.