Abstract：Intelligent learning diagnosis is a core task in smart education platforms， which evaluates and predicts learners’ knowledge status by analyzing their question-answer records to achieve personalized learning support. However， existing methods often ignore the knowledge structure information in the question-answer sequence， fail to refine the learning status to individual knowledge points， and are difficult to reveal the propagation path of knowledge mastery. To address the above shortcomings， this paper proposes an intelligent learning diagnosis method based on spatio-temporal graph convolutional network： firstly， it integrates the knowledge structure of the discipline， and treats the knowledge points as interrelated nodes； then， it uses the spatio-temporal graph convolutional network to model the temporal characteristics of the question-answer record and the spatial characteristics of the knowledge structure， and dynamically portrays the evolution of the learner’s knowledge state； finally， it generates diagnostic results with better interpretability and higher accuracy. Finally， we generate diagnostic results with better interpretability and higher accuracy. Experiments show that the diagnosis performance of the proposed method on the three real data sets is significantly better than the existing mainstream models.
2025, Vol. 59 
