An Enhanced Topic Analysis Method for Mooc Comments Based on Multi-Dimensional Feature Fusion
Chenxiao Liu · 2025
With the rapid development of MOOC education, the valuable information contained in learners' comments is crucial to improving teaching quality. However, traditional topic analysis methods often ignore the multidimensional features of comment data, resulting in inaccurate topic identification. This paper proposes an enhanced MOOC comment topic analysis method based on multidimensional feature fusion. Through an adaptive weight calculation model, it integrates features such as time decay, likes and content quality. This method introduces an improved LDA framework and optimizes the dictionary filtering mechanism. Experimental results show that compared with the baseline LDA model, our method achieves a 13 % improvement in topic coherence and nearly 30 % enhancement in perplexity. When compared to the pre-trained language model, while maintaining similar topic coherence, our approach demonstrates significantly superior topic diversity, achieving perfect diversity where the pretrained model shows limited performance. This study provides a more accurate solution for MOOC comment analysis and supports the improvement of online education quality.