Attention-Guided and Progression-Aware Neural Network Framework for Evaluating Workshop Practical Teaching Quality

From the past few years, accurate and reliable evaluation of workshop-based practical teaching quality has become difficult because of the subjective nature of feedback, various evaluation indicators, and the dynamic evolution of hands-on teaching practices. However, conventional teaching quality assessment models based on manual surveys often suffered with limited adaptability, poor interpretability and failed to capture progressive teaching enhancement. To overcome these challenges, this research proposes a Workshop-Aware Progression-Guided Artificial Neural Network (WAPANN), an improved intelligent framework for estimating workshop-based teaching quality in practice. Initially, workshop-based teaching data are gathered from professional development programs and preprocessed using data cleaning, normalization and encoding to confirm consistency. Subsequently, Black Hole Optimization (BHO) is employed to choose the most important workshop-specific practical teaching quality indicators. To overcome the challenges of static and black-box evaluation, the proposed framework presents a hands-on skill attention layer that dynamically allocates importance to hands-on teaching indicators along with a progression encoding mechanism to model temporal teaching enhancement across workshops. Furthermore, an ensemble of ANNs is created and optimized by BHO to improve prediction robustness and accuracy. Final, the teaching quality score is attained through weighted averaging of ensemble outputs, delivering adaptive and interpretable evaluation results. Experimental results demonstrated that the proposed WAPANN framework attained greater results in terms of accuracy$\mathbf{9 9. 2 8 \%}$compared to the existing ANN-BHO model.