CA-BPNN-Based Multi-Modal Fusion Teaching Quality Evaluation Method

In the context of globalization and rapid information development, enhancing educational quality has become a global focal point. Teaching quality assessment is a crucial tool for measuring educational effectiveness and a key factor in promoting teaching innovation and improving methods. Traditional assessment methods often rely on single data sources, which can lead to subjectivity and partiality. Therefore, developing a model that effectively integrates multimodal data for evaluating teaching quality is essential. This study introduces a multimodal fusion model based on CA-BPNN to address the limitations of traditional methods. The model integrates three modalities: manual scoring, online education platform data, and teaching video data. It employs a cross-attention mechanism to enhance feature extraction and uses a multimodal hybrid fusion strategy to address issues of data missingness. This approach shows significant advantages, with a reduction in mean squared error by 2.4% to 3.9% compared to traditional models. The aim of this study is to provide a more objective and comprehensive method for evaluating teaching quality, thereby aiding in the optimization of teaching management and the overall improvement of educational quality. It offers theoretical and practical guidance for educators and policymakers.