Scientometric Analysis: Trends in Deep Learning Research in the Context of Higher Education

Abstract

This study aims to map research trends regarding the application of deep learning in the context of higher education through a bibliometric approach based on Scopus data for the period 2015–2025. The analysis was conducted using the Bibliometrix package with the Biblioshiny interface in RStudio to explore various indicators, such as annual publication trends, citation distribution, contributions by country and institution, and collaboration patterns among authors. Additionally, this study utilizes visualizations such as three-field plots (Sankey diagrams), network analysis, word clouds, and tree maps to identify relationships between elements and thematic clusters in the literature. The results indicate a significant increase in the number of publications, with contributions predominantly coming from countries such as China and the United States. Citation analysis indicates an uneven distribution of scientific impact, where some articles have a more dominant influence than others. Thematically, research is developing along two main trends: a technical approach focused on the development of artificial intelligence systems and a pedagogical approach highlighting its impact on the learning process. These findings indicate that deep learning is not only positioned as a technological innovation but also as a strategy for building adaptive, personalized, and data-driven learning systems. Overall, this study provides a comprehensive overview of the direction and dynamics of global research, while also opening up opportunities for future collaboration and the advancement of research in this field.