TruthSentry: A Scalable Framework for Fake News Detection via Advanced NLP and Ensemble Learning

The proliferation of misinformation in digital media undermines trust and informed decision-making, necessitating robust detection systems. This study introduces TruthSentry, an innovative framework that leverages Natural Language Processing (NLP) and ensemble learning to identify fake news with high precision. Using a dataset of 44,898 news articles—23,481 labeled as fake and 21,417 as true—we applied comprehensive text preprocessing, including tokenization, TF-IDF, and word embeddings, followed by evaluation of five classifiers: Naive Bayes, Logistic Regression, Decision Tree, Support Vector Machine (SVM), and Random Forest. The Random Forest classifier achieved a remarkable 99.68% accuracy, demonstrating superior capability in discerning deceptive content. This work provides a high-performance, extensible solution for automated misinformation detection, with potential for enhancing digital media integrity.