Explainable Knowledge Tracing via Probabilistic Embeddings and Pattern-based Reasoning

Researchers have introduced Probabilistic Logical Knowledge Tracing (PLKT), a novel framework designed to enhance the interpretability of Knowledge Tracing (KT) models in educational technology. While existing deep learning-based KT models offer high predictive accuracy, they often suffer from opaque latent state transitions and deterministic vector embeddings, making it difficult to understand how specific past student behaviors influence predictions. PLKT addresses this limitation by employing robust Beta-distributed probabilistic embeddings to represent student knowledge states, thereby modeling the uncertainty inherent in historical learning interactions. This approach allows for explicit logical operations, such as conjunction, to construct transparent reasoning paths that clearly reveal the contribution of specific past interactions to performance predictions. Extensive experiments demonstrate that PLKT not only outperforms current state-of-the-art KT methods in terms of accuracy but also achieves superior interpretability. The study highlights a significant advancement in creating explainable AI systems for education, balancing predictive power with transparency. The associated code for the PLKT framework has been made publicly available to facilitate further research and application in the field of artificial intelligence and educational data mining.