A Bibliometric Analysis of Federated Learning: Exploring Trends In Data Privacy and Security

Studies on federated learning (FL) have gained significant traction in recent years. This approach enables the collaborative training of machine learning models without the exchange of personal data, thus overcoming the phenomenon of the so-called data islands. This study provides a rigorous bibliometric assessment of the FL landscape, specifically focusing on data privacy and security. Utilizing a curated dataset of 6,862 documents retrieved from the Scopus database, we applied performance analysis and scientific mapping techniques. We performed data visualization using biblioshiny and Python-based libraries to identify influential authors, seminal works, and key geographical and institutional contributors. Furthermore, the longitudinal mapping of thematic clusters reveals the evolution of research trends, highlighting emerging frontiers for future investigation. To the authors’ knowledge, this is the first comprehensive bibliometric study to intersect FL with data privacy and security requirements. The results suggest a shift from fundamental framework development toward specialized security applications.