面向舆论情感识别的自然语言处理技术

打开文本图片集

中图分类号：TN9173⁃34；TP399

文献标识码： A

文章编号：1004⁃373X（2025）12⁃0115⁃05

Abstract：In order to improve the accuracy and efficiency of public opinion risk prediction， a public opinion perception model based on the combination of attention mechanism and bidirectional long short⁃term memory （BiLSTM） network is proposed. This method can be used to accurately capture emotional fluctuations and contextual semantic features in public opinion data by combining the bidirectional modeling ability of BiLSTM with the feature focusing ability of attention mechanism， so as to improve the prediction accuracy of public opinion risks. By taking the "college entrance examination impersonation" incident as a sample， network public opinion data was analyzed. The effectiveness and superiority of the proposed model are verified by means of comparative experiments with various mainstream algorithms such as ELM， random forest， decision tree， LSTM， BiGRU and BiLSTM. In algorithm design， the introduction of attention mechanism can effectively improve the performance of the model in long text emotion classification， and can accurately capture key nodes of emotional changes. The experimental results show that the proposed prediction model can effectively identify public opinion risks， with an accuracy of 94.87%， which is about 5.75% higher than the best⁃performing BiGRU algorithm.

Keywords：public opinion risk identification; emotional identification; natural language processing; bidirectional long short⁃ term memory network; attention mechanism; text classification

0 引 言

随着互联网的普及和社交媒体平台的迅猛发展，公共舆论的表达方式与传播途径发生了深刻变化。（剩余6354字）