Political Sentiment Analysis on Twitter Using Deep Learning and LLM Models

In the age of rapid information spreading through social media platforms such as Twitter, Facebook, and Instagram have become serious arenas for information warfare. Coordinated misinformation campaigns and influence operations pose significant threats to societal stability and public opinion. Addressing this issue, our project combines open-source intelligence (OSINT) techniques with advanced AI-driven sentiment analysis and summarization to detect, analyses, and provide actionable insights into politically sensitive and potentially harmful narratives. The proposed system integrates a Bi-LSTM model, which achieves a high accuracy of 98%, for sentiment analysis with an LLM (LLaMA 3) to classify and interpret polarized content. This system processes large volumes of Twitter data, filtered through keywords and contextual signals, to extract and analyses politically charged or conspiratorial content. It enables interactive querying and dynamic analysis to summarize critical topics and provide meaningful insights into the detected narratives. The results demonstrate the system's capability to accurately identify sentiments, summarize critical topics, and facilitate interactive exploration, making it an effective tool for combating misinformation and enhancing situational awareness in the realm of information warfare