ARCHIVES

This project focuses on detecting fake news using a hybrid approach combining machine learning and deep learning techniques. The dataset includes real and fake news articles that are cleaned and preprocessed to improve quality. For feature extraction, TF-IDF is used to convert text into numerical form. Machine learning models such as LogisticRegression, Support Vector Machine, and Naive Bayes are trained and evaluated. Transformer-based models like BERT and RoBERTa are also implemented. Model performance is measuredusing accuracy, precision, recall, and F1-score, along with confusion matrix and ROC analysis.A majority voting technique is applied to improve final predictions. Results show that deeplearning models provide better accuracy and reliability.