Abstract:

Information can be understood as data that has already been interpreted and organized in a certain way that allows it to use it to generate of knowledge. As the volume of available information continues to grow, gaining access to what is relevant has become increasingly challenging. Scientific journals originally emerged to offer readers curated and trustworthy material, but the expansion of the internet has made it more difficult to control the quality of what any user may encounter. With the constant increase in online content, distinguishing reliable and meaningful information has turned into a demanding task for researchers and general users alike. In this context, recommendation systems offer a promising way to help readers navigate large information spaces and discover useful, trustworthy content.

Digital journals were created partly to address this need, offering users a more structured environment with verified information. Tools such as OJS (Open Journal Systems) support the entire editorial workflow and, thanks to their modular and open-source nature, have been widely adopted by academic publications around the world.

Despite its advantages, OJS includes only basic recommendation mechanisms with limited personalization. To address this gap, we propose a hybrid recommendation system that integrates the main filtering strategies and employs modern technologies to tailor suggestions to different user profiles. This work describes the content-based and collaborative approaches used, the architecture of the developed module, its implementation, and the evaluation performed using standard metrics for recommendation systems.

The system was evaluted using Abstraction & Application, a journal published by the Faculty of Mathematics at the Autonomous University of Yucatán, a leading institution in southeastern Mexico. As the journal’s collection of articles in mathematics and informatics has continued to grow, it has become increasingly difficult for readers to locate material aligned with their interests, making it a suitable context for this study. The system’s performance was measured using widely adopted evaluation metrics, including accuracy and recall.

Our initial results show significant improvements compared with similar systems created for OJS. The main contribution of this work lies in demonstrating that OJS can be extended through external components without modifying its core, allowing scientific journals to adopt advanced recommendation systems with minimal technical intervention.

Keywords:

Recommendation Systems, hybrid filtering, Open Journal Systems (OJS), scientific publishing.