A Multi-Channel Text Sentiment Analysis Model Integrating Pre-training Mechanism

• DOI: 10.5755/j01.itc.52.2.31803
• Publication Date: 2023-07-15T04:40:25Z
• Authors: Shengbin Liang; Jiangyong Jin; Wencai Du; Shenming Qu
• Keywords: Softmax function; Sentiment Analysis

The number of tourist attractions reviews, travel notes and other texts has grown exponentially in the Internet age. Effectively mining users’ potential opinions emotions on attractions, helping to provide users with better recommendation services, which is great practical significance. This paper proposes a multi-channel neural network model called Pre-BiLSTM combined pre-training mechanism. uses combination coarse fine- granularity strategies extract features text information such as reviews improve performance sentiment analysis. First, we construct three channels use improved BERT skip-gram methods negative sampling vectorize word-level vocabulary-level text, respectively, so obtain more abundant textual information. Second, mechanism generate deep bidirectional language representation relationships. Third, vectors are input into BiLSTM parallel global local features. Finally, fuses classifies them using SoftMax classifier. Furthermore, numerical experiments conducted demonstrate that outperforms baselines by 6.27%, 12.83% 18.12% average terms accuracy, precision F1-score.