Ideological and Political Education (IPE) aims to shape learners’ political understanding, moral awareness, and social responsibility. Text analysis enhances this process by automating the evaluation of large-scale educational content, enabling more consistent, efficient, and data-driven insights. Advanced Natural Language Processing (NLP) techniques play a crucial role in interpreting complex ideological texts and improving the quality of educational feedback. In this study, we address existing limitations by integrating two advanced transformer-based models—RoBERTa-GRU for deep contextual classification and GPT-4 for higher-order semantic verification. RoBERTa-GRU leverages robust attention mechanisms and sequential modeling to capture long-range dependencies, while GPT-4 provides reasoning-based validation to enhance reliability. Model performance is assessed using accuracy, precision, recall, and F1-score. Experimental results demonstrate that the proposed RoBERTa-GRU model achieves a 93.5% F1-score, 85% recall, and 76.37% accuracy on Dataset 1, outperforming baselines such as CNN (74.83%) and BiLSTM-BERT (74.66%). On Dataset 2, the model achieves 65.57% accuracy, surpassing traditional RNN and transformer counterparts.
Lichao Wang (Wed,) studied this question.