Campus public opinion monitoring method based on Bi-LSTM and attention mechanism

Abstract

With the widespread use of social media on campus, online public opinion has become a major factor affecting the stability of universities. Traditional public opinion monitoring methods mainly rely on keyword filtering and statistical models. These methods fail to capture the contextual semantic features of text and lack the ability to predict the dynamic evolution of sudden opinion events. To address these limitations, this paper proposes a campus online public opinion monitoring model that combines Bidirectional Long Short-Term Memory and Attention Mechanism. The Bidirectional Long Short-Term Memory learns the temporal features of text in both directions, enhancing its ability to represent contextual information. The Attention Mechanism dynamically focuses on key opinion-related words, improving the efficiency of extracting critical information. Experimental results show that the model achieved an F1 score of 0.81 and a recall of 0.72 on the THUCNews dataset, slightly lower than BERT’s 0.73. On the Twitter dataset, the recall was 0.75. In sentiment analysis, the model achieved a strength score of 0.82 on 10,000 data points, 0.07 higher than the second-best model. In terms of text analysis efficiency, the proposed model achieves 78.35% on the THUCNews dataset and 82.47% on the Twitter dataset, both higher than the three comparative models. These results demonstrate that the proposed model has high accuracy and applicability in sentiment classification of campus public opinion. It provides a new approach for public opinion monitoring on campus and supports its development toward efficient systems.