Real-Time Integrated Social Media Feedback Analysis System

This research presents a Real-time Integrated Social Media Feedback Analysis System that extracts, analyzes, and summarizes user opinions from Reddit and YouTube. With rapid growth of user-generated content on these platforms, understanding public sentiment supports decision-making in marketing, product development, and social analysis. The system collects data through official and public APIs, processes unstructured text, and applies the VADER (Valence Aware Dictionary and sentiment Reasoner) lexicon model for positive, neutral, and negative classification—well suited to informal social text. For interpretability, the system integrates hosted large language models via the Groq API to produce concise summaries of collected feedback. The architecture uses a React frontend and Flask backend for interactive dashboards (Chart.js visualizations), scalable separation of concerns, and deployable REST endpoints. Query relevance filtering, duplicate removal, and HTML entity normalization improve the quality of retrieved posts and comments. Under typical API conditions, end-to-end analysis completes in approximately 3–8 seconds for representative queries. Limitations include sarcasm and irony, API rate limits, English-oriented filtering, and sampling bias across platforms. The work demonstrates a practical intelligent systems pipeline from ingestion to visualization.