Detecting Fake News Using Deep Learning and Text Classification

Abstract The exponential growth of digital media has amplified the spread of fake news, posing significant challenges to information authenticity and public trust. This research presents an automated framework for fake news detection using Deep Learning and Text Classification techniques. The study leverages Natural Language Processing (NLP) to extract linguistic, semantic, and contextual features from news articles. Multiple deep learning architectures— Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), Transformer, and a Hybrid CNN-LSTM model—were implemented and evaluated on the Kaggle Fake and Real News Dataset. Comprehensive preprocessing, including tokenization, stopword removal, and lemmatization, was applied to enhance data consistency. Model performance was assessed using standard evaluation metrics such as accuracy, precision, recall, and F1-score. Experimental results demonstrate that the Hybrid CNN-LSTM model achieved the highest accuracy of 95.70%, outperforming individual models by effectively capturing both local and sequential text patterns. The findings confirm that combining convolutional and recurrent learning approaches enhances classification robustness. This study contributes to the development of scalable, automated, and reliable tools for misinformation detection, thereby supporting media integrity and reducing the societal impact of false information dissemination.