Multi-Level Online Public Opinion Sentiment Analysis Method Based on Text Features
Experimental results show that BBMC achieves better performance than the evaluated baseline models in terms of accuracy, recall, and F1-score, indicating the empirical effectiveness of the proposed task-oriented feature-integration strategy and capsule-based classification head for online public opinion sentiment analysis.