Enhanced digital library with book recommendations based on collaborative filtering

Abstract

This research aims to develop a digital library system by implementing a recommendation system becomes a feature of the digital library system. The main problem with the digital library system is the large number of books that users will have difficulties finding interesting books. Therefore, Machine Learning is applied using the User-based Collaborative Filtering method to provide book recommendations for users. The book recommendation system being developed not only uses book value as input but also user behavior and book borrowing data. The resulting recommendations are divided into three parts according to the input which are book value, user behavior, and book borrowing data. The system developed is web-based using the PHP programming language, with modifications to the available open-source digital library system. Experiments are carried out by testing the system to university students and satisfaction questionnaire to the students. The total result of the average precision-recall calculation has an accuracy of 79% from 35 data users which includes an assessment of 1000 types of books with different categories. It means that books recommended by the system are relevant to users. The developed models and applications are potentially used as smart library applications.