Abstract:
The study of understanding emotions and sentiments present in textual data is the focus of the developing area of emotion and sentiment analysis. The inability of modern technology to precisely recognize and comprehend complicated human emotions, despite tremendous breakthroughs, as demonstrated by instances such as Microsoft’s ”Tay.ai” bot, which produced offensive content and was unable to discern human moods. These difficulties arise from the complexity of human language, particularly when it comes to interpreting complex and situation-specific emotional responses. The limited capacity of traditional machine learning models, such Support Vector Machines (SVMs), to capture temporal correlations in sequential data makes them difficult to handle these complications. To overcome these constraints, this study investigates a hybrid model that combines SVMs (renowned for their great classification performance in high-dimensional environments) with Gated Recurrent Unit (GRU)-based Recurrent Neural Networks (RNNs), which excel at capturing temporal correlations in sequence data. The proposed SVM-GRU hybrid model seeks to increase sentiment analysis accuracy (up to 84.88%) by efficiently managing both high-dimensional features and text’s sequential character. Extensive testing revealed that the hybrid model outperformed solo models, particularly in terms of detecting subtle emotional cues and retaining contextual coherence. A carefully selected dataset was utilized to assure the model’s applicability to a wide range of real-world scenarios. This study emphasizes hybrid models’ potential for improving sentiment analysis accuracy, as well as their practical value in industries such as healthcare, customer service, marketing, finance, and political analysis. Future work will focus on enhancing the model’s scalability and real-time processing capabilities, therefore contributing to the creation of AI systems capable of recognizing and responding to human emotions with more sensitivity and precision.
