EduRiskX Combines Transformers and F-Logic for Early Academic Risk Prediction

Summary

EduRiskX is a neuro-symbolic framework for predicting academic risk in online education. It combines a temporal Transformer, an F-Logic rule base grounded in educational theories, and logistic-regression fusion. On the OULAD dataset, the system reached 0.900 accuracy, a 0.894 F1-score, an average detection point of Week 9.32, and a 94.30% detection rate. It outperformed several time-series and deep-learning baselines while producing structured explanations tied to observable student behavior.