Hybrid sentiment analysis on Twitter data using VADER and RoBERTa models

Twitter and other social media platforms generate large volumes of user-generated text that reflect public opinion in real time. However, sentiment analysis of Twitter data remains challenging due to informal language, abbreviations, contextual dependencies, and the frequent presence of implicit sentiment. Lexicon-based approaches such as VADER provide computational efficiency but often fail to capture contextual meaning, while transformer-based models such as RoBERTa achieve higher accuracy at the cost of increased computational requirements. This paper presents a hybrid sentiment analysis framework that integrates VADER and RoBERTa to balance accuracy and efficiency. The proposed system is implemented as a Flask-based web application with MySQL database support, enabling user authentication, keyword-based tweet filtering, and result visualization. GPU acceleration and batch processing techniques are employed to optimize performance. Experimental evaluation conducted on a dataset of 1.6 million tweets demonstrates that the hybrid approach achieves an accuracy of 89.1%, outperforming standalone lexicon-based methods. Statistical analysis confirms that the observed improvements are significant (p < 0.001). The results indicate that the proposed framework is suitable for large-scale sentiment analysis applications, including brand monitoring, political opinion analysis, and market trend assessment.