SENSYS: A MULTIMODEL SENTIMENT ANALYSIS ON SOCIAL MEDIA DATA

Social media platforms generate vast amounts of multimodal content, including text, images, and videos, each conveying rich emotional information; however, most existing sentiment analysis methods focus primarily on text, overlooking critical context from visual and audiovisual data. This paper presents SENSYS (Sentiment Analysis System), a multimodal framework designed to analyze sentiment across text, image, and video content within a unified pipeline. The system operates on a simulated social media environment populated by AI-driven bot users, enabling scalable and privacy-preserving experimentation. SENSYS integrates BLIP for visual caption generation and the Groq-hosted LLaMA 3.3 70B model for sentiment classification, while a React.js-based dashboard provides real-time visualization of sentiment trends. Experimental evaluation demonstrates an overall accuracy of 88.4%, with text, image, and video achieving 91.2%, 87.6%, and 84.3%, respectively, ensuring reproducibility, scalability, and effective multimodal sentiment monitoring.