Global Local Fusion Neural Network for Multimodal Sentiment Analysis

With the popularity of social networking services, people are increasingly inclined to share their opinions and feelings on networks, leading rapid increase in multimodal posts various platforms. Therefore, sentiment analysis has become a crucial research field for exploring users’ emotions. The complex complementary interactions between images text greatly heighten difficulty analysis. Previous works conducted rough fusion operations ignored study fine features task, which did not obtain sufficient interactive information This paper proposes global local neural network (GLFN), comprehensively considers features, aggregating these analyze user sentiment. model first extracts overall by attention modules as modality-based features. Then, coarse-to-fine learning is applied build effectively. Specifically, cross-modal module used fusion, fine-grained capture interaction objects words. Finally, we integrate all achieve more reliable prediction. Extensive experimental results, comparisons, visualization public datasets demonstrate effectiveness proposed classification.