A hybrid book recommendation model using deep learning, collaborative, and content filtering

Abstract

The continuous evolution of technology is transforming the way libraries interact with their users, and in turn, how users engage with books. Recommendation systems are conceived as information filtering systems whose goal is to provide access to personalized information (books of interest, magazines, databases, scientific articles, rooms, etc.) to enhance the user experience, promote the usability of bibliographic resources, and optimize services. This article proposes a hybrid model for the automatic recommendation of books that combines three processes in two phases: user identification, collaborative filtering, and content filtering. In the first phase, the user recognition process is carried out using techniques that implement deep learning, and in the second phase, the recommendation processes are integrated through collaborative filtering and content filtering. A case study was developed in a library environment for recommending books and was evaluated using classic information retrieval metrics. The results were compared with other more robust recommendation models, obtaining satisfactory outcomes.