Leveraging a graph attention network-based academic recommendation framework for Indian higher education institutions

Ramachandra H V; Dr. Biradar Shilpa

doi:10.37965/jait.2026.0916

Leveraging a graph attention network-based academic recommendation framework for Indian higher education institutions

Authors

Ramachandra H V Research Scholar, Department of Computer Science and Engineering, RNS Institute of Technology, Bengaluru – 560098, Affiliated to Visvesvaraya Technological University, Belagavi - 590018, Karnataka, India; Assistant Professor, School of Computer Science and Engineering, REVA University, Bengaluru – 560064, Karnataka, India https://orcid.org/0000-0002-1411-9067
Dr. Biradar Shilpa Department of CSE(DS), RNS Institute of Technology, Bengaluru – 560098, Affiliated to Visvesvaraya Technological University, Belagavi – 590018, Karnataka, India

DOI:

https://doi.org/10.37965/jait.2026.0916

Keywords:

Artificial intelligence (AI), educational data mining, graph attention network (GAT), graph neural networks (GNNs), higher education admissions, recommender system

Abstract

The competitive nature of admission to the top Indian institutes on the basis of the Graduate Aptitude Test in Engineering (GATE) requires intelligent systems in the student–college matching. Conventional recommendation procedures usually do not take into consideration important aspects like cutoff differences and institutional ratings, resulting in minimal impartiality and transparency to the students. Following this approach, this paper suggests a recommendation framework using the graph attention network (GAT), where students and colleges are treated as graph nodes, and compatibility as an edge is represented by GATE scores, category cutoffs, course preferences, and seat availability. The attention mechanism allows adaptive learning of feature relevance, improving the quality of personalized suggestions. The system produces optimal-3 college suggestions to each student and is tested with various ranking and error-based measurements to insist on stability. Experimental analysis underlines the promise of GAT-based models to provide fair, clear, and data-driven solutions, helping students make informed choices and make a contribution to more efficient admission of higher education.