New AI Recommends Research Collaborators and Explains Why

A new explainable scholar recommendation model called PAGSL aims to identify potential scholars of interest and provide reasonable explanations for recommendations. It combines representation learning and semantic path knowledge, activates latent academic paths through graph augmentation, and uses a path-based neural network plus a masking mechanism with path score optimization to generate concise, informative explanations. On three academic datasets—Scholat, AMiner, and ResearchGate—comprising about 49,000 authors and over 330,000 interaction records, PAGSL outperformed state-of-the-art baselines with improvements of 6.31% in F1-score, 6.29% in Accuracy, and 5.95% in AUC for recommendation, and it also provided quantitative results and visualization effects for explainability.