An improved RCBT model and its application to sentiment analysis of movie comments

Sentiment analysis of social media comments is crucial for understanding public opinion and trends. However, existing models like Bidirectional Encoder Representations from Transformers-Bidirectional Long Short-Term Memory (BERT-BiLSTM) face challenges with long texts and complex sentiment expressions. To address these issues, this paper proposes an improved RoBERTa-CNN-BiLSTM-Transformers (RCBT) model for sentiment analysis of social media comments. Specifically, a CNN module is introduced to enhance the model's capability in extracting local features, efficiently capturing local patterns such as phrases and word combinations, thereby improving sentiment analysis accuracy. Additionally, the Transformers mechanism is incorporated to address the high computational complexity and low training efficiency of BiLSTM in handling long-range dependencies. The self-attention mechanism in Transformers allows for better capture of long-range dependencies without significantly increasing computational complexity, thus enhancing the model's performance in long text sentiment analysis. Finally, RoBERTa is employed for text feature extraction and representation. Compared to BERT, RoBERTa has stronger pre-training capabilities and can handle larger datasets, offering more accurate predictions in sentiment analysis tasks. The proposed RCBT model is validated on the IMDB movie review dataset. Experimental results show that the model performs excellently in classification tasks, with precision (P), recall (R), and accuracy (ACC) reaching 99.80% on the training set, fully validating the model's effectiveness. Compared to other models, the proposed model achieves the highest accuracy (ACC) and F1 score on the test set, confirming its superiority.