Neuro-Symbolic EduRiskX Framework Combines Transformers and F-Logic for Interpretable Academic Risk Prediction
Researchers have introduced EduRiskX, a neuro-symbolic framework that integrates temporal Transformer models with F-Logic symbolic reasoning to predict academic risk early and interpretably.
A new research paper published on arXiv introduces EduRiskX, a neuro-symbolic framework designed to predict academic risk in online learning environments. Traditional deep learning models in educational technology often operate as black boxes, making it difficult for educators to trust and act on their predictions. EduRiskX addresses this limitation by blending the predictive power of neural networks with the explainability of symbolic logic.
Primary source
Open arXiv ↗This article summarizes the source. Check the original for full details and conditions.
Comparison
| Aspect | Before / Alternative | After / This |
|---|---|---|
| Architecture | Purely deep learning models (e.g., PatchTST, iTransformer) acting as black boxes | Neuro-symbolic framework combining temporal Transformers with F-Logic reasoning rules |
| Interpretability | Low explainability, leading to a trust crisis in real-world educational deployment | High explainability via a symbolic rule base grounded in pedagogical theories |
| Risk Fusion | Manual thresholding or single-model end-to-end predictions | Logistic regression-based fusion of neural risk probability and symbolic confidence |
| Decision Rule Source | Fully learned statistical weights without theoretical alignment | F-Logic rules extracted from training data using Engagement and Student Integration models |
