As the web continuously evolves, the information quantities online grow progressively. Therefore, the automatic processing of massive online data has become a research hotspot. Text classification serves as an effective method for analyzing unstructured web data. This study focuses on employing deep learning techniques for sentiment classification of textual information in social networks, aiming to develop an automated sentiment analysis tool. The primary objective is to analyze and process massive volumes of unstructured online data while overcoming the limitations of traditional text analysis methods, such as statistical or machine learning methods which heavily depend on manually engineered features. This paper constructs a sentiment classification model based on a Bidirectional Long Short-Term Memory network (Bi-LSTM) integrated with an attention mechanism, which classifies user comments into positive or negative categories. Experimental results demonstrate that, compared with single-layer LSTM, Bi-LSTM, and two-layer Bi-LSTM models, the proposed model achieves more accurate text classification performance.
W. Ma (Wed,) studied this question.