As the number of digital resources expands and expands it becomes increasingly difficult to recommend reading matter. Research in the International Journal of Information and Communication Technology has led to a new artificial intelligence (AI) system that improve on precision and variety of book recommendations for online library goers. The new approach blends two established techniques, content-based filtering (CBF) and collaborative filtering (CF), and then has its roots in an advanced machine learning algorithm – Extreme Learning Machine (ELM) – which allows it to come up with the perfect personalized recommendation for the reader.
In traditional recommendation systems, Tianhao Wu of Changchun University of Technology, China explains, content-based filtering suggests books based on a book’s attributes, such as its title, author, and genre. Collaborative filtering by contrast makes recommendations based on user behaviour, what books they have read previously and how they rated them. By combining both systems with ELM, the new hybrid model aims to improve the accuracy of suggestions while also increasing their diversity, better reflecting a user’s unique preferences and opening them up to new books they may not have encountered otherwise but will hopefully enjoy.
ELM can process large datasets quickly and efficiently, which is particularly useful in the context of online libraries, where both the number of books and user interactions can be immense. Unlike traditional neural networks, ELM reduces the complexity of training by randomly generating weights for each entry. This allows it to adapt to new data much more quickly than other approaches and with greater accuracy.
As digital libraries continue to grow, this new hybrid system holds the potential to transform how books are recommended to users, making their library experiences more personalized and efficient. The team will attempt to address remaining challenges such as the cold-start problem facing new users about which there is initially no reading experience data and similarly with new books from new authors that equally lack a data history.
Wu, T. (2025) ‘An ELM-based approach to promoting reading of library books’, Int. J. Information and Communication Technology, Vol. 26, No. 2, pp.82–95.