RAG-KT: Cross-platform Explainable Knowledge Tracing with Multi-view Fusion Retrieval Generation
Summary: arXiv:2604.10960v1 Announce Type: new
Abstract
Knowledge Tracing (KT) infers a student’s knowledge state from past interactions to predict future performance. Conventional Deep Learning (DL)-based KT models are typically tied to platform-specific identifiers and latent representations, making them hard to transfer and interpret. Large Language Model (LLM)-based methods can be either ungrounded under prompting or overly domain-dependent under fine-tuning. In addition, most existing KT methods are developed and evaluated under a same-distribution assumption. In real deployments, educational data often arise from heterogeneous platforms with substantial distribution shift, which often degrades generalization.
Introduction to RAG-KT
To address these challenges, we propose RAG-KT, a retrieval-augmented paradigm that frames cross-platform KT as reliable context constrained inference with LLMs. This innovative approach builds a unified multi-source structured context with cross-source alignment via Question Group abstractions. By retrieving complementary rich and reliable context for each prediction, RAG-KT enables grounded prediction and interpretable diagnosis.
Key Features of RAG-KT
- Cross-platform Capability: RAG-KT effectively integrates data from various educational platforms, ensuring that the model is adaptable and robust.
- Grounded Predictions: The model leverages rich context to provide predictions that are not only accurate but also interpretable, allowing educators to understand student performance better.
- Question Group Abstraction: This feature allows the model to align and contextualize data from diverse sources, enhancing the quality of predictions.
- Enhanced Generalization: By addressing the distribution shift commonly found in real-world data, RAG-KT improves the model’s ability to generalize across different educational contexts.
Experimental Validation
We conducted experiments on three public KT benchmarks to validate the effectiveness of RAG-KT. The results demonstrate consistent gains in accuracy and robustness, particularly under cross-platform conditions. Our findings indicate that RAG-KT not only outperforms conventional KT models but also provides a framework for future research in knowledge tracing.
Conclusion
RAG-KT represents a significant advancement in the field of knowledge tracing, particularly in its ability to handle heterogeneous educational data across platforms. By utilizing a retrieval-augmented approach, it offers a promising direction for developing more interpretable and effective educational technologies. As the landscape of educational data continues to evolve, RAG-KT stands out as a robust solution for the challenges faced in knowledge tracing.