Context-Aware Sentiment Analysis for Enhanced Fake News Detection

The extensive use of social networking sites has transformed how people share and consume information, but it has also made it easier for fake news to proliferate quickly, endangering social stability and public confidence. This study presents an innovative framework for detecting fake news by integrating context-aware sentiment analysis with advanced machine learning techniques. The approach uses contextual embeddings and sentiment polarity scores to identify news stories as either real or false, utilizing the Indian Fake News Dataset. The effectiveness of several machine learning and deep learning models, such as Decision Trees, Gradient Boosting, BERT, BART, and Logistic Regression, in identifying false news was assessed. Among these, deep learning models, particularly BERT, achieved the highest accuracy at 92%, significantly outperforming traditional models such as Logistic Regression (75%) and Gradient Boosting (85%). The findings emphasize the effectiveness of combining context-aware sentiment analysis with state-of-the-art deep learning models, positioning BERT as a robust and scalable tool for combating misinformation in real-time digital ecosystems.