EduRiskX: A Neuro-Symbolic Framework with F-Logic Reasoning for Early Academic Risk Prediction

Predicting students’ academic risk in online education is important for enabling timely educational intervention. However, many existing models exhibit limited early detection capability and insufficient interpretability to support pedagogi- cal decision-making. To address these challenges, this paper proposes EduRiskX, a neuro-symbolic framework that integrates a temporal Transformer-based predictor with F-Logic symbolic reasoning. The neural component models longitudinal student activity sequences using tem- poral attention, class-weighted loss, and dynamic weekly truncation. The neural predictor is first trained independently. Subsequently, an F-Logic rule base is constructed exclusively from the 80% training set and grounded in establishededucational constructs. The neural risk probability and symbolic confidence score are then combined through a logistic regression—based fusion mechanism, allow- ing the relative contribution of neural and symbolic signals to be learned in a data-driven manner. Experiments conducted on the Open University Learning Analytics Dataset (OULAD) using a strict 80/10/10 student-level split show that EduRiskX achieves an accuracy of 0.900 and an F1-score of 0.894 at the end of the semester (Week 38), with an average early detection week of 9.32 and a detection rate of 94.30%. Compared with state-of-the-art (SOTA) time-series models, including PatchTST and iTransformer, as well as commonly used deep learning base-lines such as LSTM and CNN, EduRiskX yields improved recall and earlier risk identification under identical experimental conditions. In addition to predictiveperformance, the F-Logic module provides structured rule-based explanationslinking risk predictions to observable behavioral patterns, supporting transparent and pedagogically aligned early intervention in online learning environments.