<![CDATA[Deep learning has emerged as a critical pedagogical framework integrating cognitive depth with character development in elementary education, yet systematic evidence mapping research trends remains limited. This study analyzes trends in deep learning research for elementary school character building during 2020-2025. A systematic review following PRISMA methodology examined 690 publications from Scopus and Google Scholar databases. Following rigorous screening using predetermined inclusion-exclusion criteria, bibliometric analysis was conducted using VOSviewer software to identify publication patterns, geographic distribution, methodological approaches, and thematic evolution. Analysis revealed exponential publication growth from 2020 to 2025, peaking in 2024-2025, with Indonesia dominating (65% of publications). Research and Development methodology predominated (46%), while questionnaires and tests dominated assessment approaches (89%). Network visualization demonstrated paradigm shift from technology-oriented computational approaches toward humanistic pedagogies emphasizing meaningful understanding and values integration. Temporal analysis revealed emerging themes including project-based learning, moral education, and Pancasila student profile dimensions, reflecting alignment with Indonesia's Merdeka Curriculum implementation. Findings document significant conceptual transformation wherein "deep learning" terminology migrated from artificial intelligence to pedagogical domains. However, methodological limitations persist, including limited longitudinal designs, cross-cultural comparisons, and authentic character assessment instruments. Results illuminate research-policy dynamics while identifying priorities for advancing theoretical understanding and practical implementation of character-integrated deep learning pedagogies in elementary education.]]>
Copyrights © 2026
