Deep Learning-Based Multi-Modal Sentiment Analysis and Research

With the rise of social media and online platforms, the need for comprehensive analysis of multimodal data such as text, images, and audio has been increasing, making the effective fusion of these data crucial for the accuracy of sentiment analysis. However, existing multimodal sentiment analysis models often face issues with insufficient precision and low processing efficiency when dealing with complex data. To address these challenges, this paper introduces a new multimodal sentiment analysis model, the BLR model, which integrates Transformer and Bi-LSTM technologies to optimize the feature fusion process. Tests on standard datasets combined with the MVSA sub-datasetshow that the BLR framework enhances classification precision by 3.025% and 2.875% respectively, relative to conventional baselines. Moreover, ablation experiments confirm the BLR model's advantages in maintaining data integrity and enhancing processing efficiency, particularly in complex multimodal data handling tasks. These results not only demonstrate the advanced nature of the BLR model but also provide a new direction for future research in multimodal sentiment analysis.