Intelligent Generation of Physical Education Teaching Video Assessment by Integrating Big Language Modeling and Stacking Algorithm

At present, the traditional physical education teaching mode can no longer meet the ever-changing teaching changes, and how to improve the quality of physical education teaching has become the focus of current research. To improve the quality of physical education teaching and enhance students' mastery of physical education knowledge points, a method for intelligent generation of physical education teaching video evaluation based on the fusion of big language model and stacking algorithm is proposed. The new method extracts semantics from physical education teaching videos through a large language model, and uses a stacking algorithm to evaluate and analyze the generated test questions. The research results indicate that the stacked algorithm model has significantly higher indicator values than traditional models in the evaluation of short answer and essay questions. After using the model in teaching validation analysis, the number of knowledge points mastered by students increased from 8 to 9, and the evaluation score increased from 4.2 points to 8.5 points. It can be seen that the new method can significantly improve the evaluation effect of sports teaching videos and increase students' mastery of teaching knowledge points. This has good guiding value for improving the quality of physical education teaching.