Scholarlink

<p>This study investigates the innovative use of deep learning models in ideological and political education (IPE) at vocational colleges. The study focuses on addressing two core challenges in traditional IPE: limited adaptability of educational resources and low student engagement. Using datasets related to resource allocation and learner performance, the study applies Graph Neural Networks (GNNs) and a Multimodal Meta-Learning Frameworks (MMLFs). These are combined with the extended sequence modeling capabilities of Transformer-XL to build a dynamic resource optimization model. The model incorporates 42 features spanning three dimensions—learner characteristics, resource attributes, and environmental factors—through heterogeneous data fusion. A multi-head attention mechanism enables cross-feature interaction, while a curriculum knowledge graph maps resources to specific competencies. Experimental results show that the model achieves a resource effectiveness prediction accuracy of 89.7%, surpassing traditional methods by 23.5%. It also improves knowledge acquisition by 37.2% and raises the positive behavioral transformation rate by 41.3%. Course completion rates increased to 0.87, and cross-cultural transfer tests maintained an accuracy of 83.4%. Furthermore, the dynamic optimization mechanism reduced resource redundancy by 32% and improved teacher management efficiency by 80%. Demonstrating strong robustness in cross-cultural educational contexts, this model offers a promising pathway for transforming IPE through artificial intelligence.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

The impact of AI technology for ideological and political education teaching based on deep learning

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月1日

查看全文
  • Yongjian Wang

摘要

This study investigates the innovative use of deep learning models in ideological and political education (IPE) at vocational colleges. The study focuses on addressing two core challenges in traditional IPE: limited adaptability of educational resources and low student engagement. Using datasets related to resource allocation and learner performance, the study applies Graph Neural Networks (GNNs) and a Multimodal Meta-Learning Frameworks (MMLFs). These are combined with the extended sequence modeling capabilities of Transformer-XL to build a dynamic resource optimization model. The model incorporates 42 features spanning three dimensions—learner characteristics, resource attributes, and environmental factors—through heterogeneous data fusion. A multi-head attention mechanism enables cross-feature interaction, while a curriculum knowledge graph maps resources to specific competencies. Experimental results show that the model achieves a resource effectiveness prediction accuracy of 89.7%, surpassing traditional methods by 23.5%. It also improves knowledge acquisition by 37.2% and raises the positive behavioral transformation rate by 41.3%. Course completion rates increased to 0.87, and cross-cultural transfer tests maintained an accuracy of 83.4%. Furthermore, the dynamic optimization mechanism reduced resource redundancy by 32% and improved teacher management efficiency by 80%. Demonstrating strong robustness in cross-cultural educational contexts, this model offers a promising pathway for transforming IPE through artificial intelligence.