Big Data-Driven Innovation in University Ceramic Education and Teaching Practices

Pottery education often relies on subjective feedback and delayed evaluations, which can hinder students’ skill development and creative expression. Traditional methods lack the ability to provide real-time, personalized guidance, which is critical in hands-on learning environments. To address these limitations, we propose a novel system based on a Spatio-Temporal Graph Convolutional Network (ST-GCN) that integrates an attention mechanism, real-time feedback, and adaptive learning. The ST-GCN model is designed to monitor students’ behavior during pottery-making, providing immediate feedback to improve both technical skills and creativity. The model’s key advantages include its ability to capture spatio-temporal relationships in student actions, dynamically prioritize important movements, and tailor task difficulty based on performance. Experimental results show that the proposed ST-GCN model outperforms traditional models such as RNN-LSTM and CNN-based approaches in key metrics like task completion time, skill improvement, and feedback effectiveness. The ablation study further demonstrates the importance of each model component, with significant performance declines when key components are removed. This research provides an efficient, data-driven solution for pottery education, offering real-time, individualized feedback that enhances both technical proficiency and creative exploration. The proposed model represents a meaningful step toward optimizing the learning process in art education through advanced machine learning techniques.