Detection of Fake News by Convolutional Neural Networks and Recurrent Neural Networks

The proliferation of fake news on digital platforms poses significant threats to public trust, democracy, and social stability. In this study, a hybrid model by agglomerating the two different neural networks has been introduced for detecting fake news accurately. CNN captures local textual patterns, while the RNN models temporal dependencies for enhanced context understanding. Key data preprocessing techniques such as HTML tag removal, URL elimination, stopword removal, and the use of GloVe word embeddings were applied to ensure data integrity. The proposed model achieved high performance, with a testing accuracy of 91.80%, an F1-score of 0.91, and an Area Under the Curve (AUC) of 0.97, demonstrating its robustness in differentiating news. These results suggest that the hybrid architecture effectively leverages both local and sequential features for improved fake news classification.