Empowering Fake News Detection Through Innovative Hybrid Deep Learning-Based Approach

The essence of getting information to the populace in the shortest time possible through social media and other interactive technologies has made it easier for fake news to thrive hence creating a menace to any society as well as its decision-making systems. This research work presents a sound AI-based approach to classify fake news using DL algorithms embedded within a CNN-BiLSTM model. Using the ISOT dataset, the comparison of the outcomes achieved by the proposed model with the original models shows that the proposed model has the highest accuracy score of 94.07%, a precision of 93.50%, a recall of 94.60%, and an F1-score of 94.00%. The proposed hybrid model architecture of CNN for spatial feature extraction and BiLSTM for capturing contextual dependencies that are essential for text analysis and have relatively high accuracy compared to conventional ML algorithms, including CNN, RNN, LSTM, and other DL methods. The proposed approach is verified by a set of comprehensive measures, including ROC curve analysis and various training-validation metrics that reveal the effectiveness of the presented approach in overcoming the challenges of fake news identification.