Abstract

We propose a hybrid e-book recommendation mechanism that leverages collaborative filtering and contentbased recommendation paradigms to address inherent challenges in e-learning systems. For collaborative filtering, we present an innovative deep learning framework that utilizes embeddings to enhance accuracy and manage large datasets efficiently. This framework effectively addresses the cold start problem, thereby improving recommendation precision. In content-based recommendation, we introduce a regression-based technique to elevate system capabilities by incorporating content attributes. The integration of these techniques into our deep learning model creates a comprehensive and adaptable solution with scalability and effectiveness. Experiments on the Book Recommendation dataset demonstrate that our solution provides better suggestions and outperforms existing works in terms of Root Mean Square Error (RMSE) and Mean Absolute Error (MAE), achieving values of 0.69 and 0.51, respectively

Information

Item Type:	Journal
Divisions:	» Laboratory of Computer Science and Mathematics
ePrint ID:	5300
Date Deposited:	2025-02-07
Further Information:	Google Scholar
URI:	https://www.univ-soukahras.dz/en/publication/article/5300

BibTex