EXPLAINABLE AI IN HIGHER EDUCATION ASSESSMENT: A BIBLIOMETRIC EXPLORATION OF AN EMERGING RESEARCH DOMAIN
Abstract
This study employs bibliometric analysis following GLOBAL recommendations to systematically map the emerging research landscape on explainable artificial intelligence (XAI) in higher education assessment, pursuing five objectives: quantifying research productivity, assessing geographic distribution, identifying collaborative networks, uncovering thematic evolution, and mapping intellectual foundations. We analyzed 72 peer-reviewed publications from Scopus spanning 2019-2025 to address a critical gap in systematic knowledge about AI transparency in educational contexts, particularly in response to regulatory frameworks such as the GDPR and the EU AI Act. The analysis reveals accelerating research activity, with publication output growing from single contributions in 2019-2020 to 26 documents in 2024 (R² = 0.7666). Geographic distribution shows China leading in productivity (8 documents) while Sweden demonstrates exceptional citation efficiency (25.50 citations per document), with European institutions collectively accounting for 35.4% of contributions. The field generates moderate but growing scholarly impact with 655 total citations, achieving an h-index of 13 and g-index of 24. Science mapping identifies four primary thematic domains: AI transparency methodologies, educational analytics applications, machine learning implementations, and student performance prediction systems. Author collaboration analysis reveals Dragan Gasevic as the most productive researcher (5 documents, 61 citations) alongside a highly integrated four-author cluster (Afzaal, Fors, Nouri, and Zia) achieving exceptional impact with 101 citations each from 3 joint publications. Cocitation analysis identifies Adadi and Berrada’s XAI survey (2018) as the primary theoretical anchor. The temporal evolution indicates progression from fundamental XAI concepts toward practical educational applications, with innovative South-South partnerships emerging among developing economy institutions. The findings provide systematic preliminary evidence for growing academic attention to AI transparency in educational contexts, highlighting critical needs for terminological standardization and longitudinal investigation of regulatory drivers shaping this emerging field.