A Data-Driven Advanced Deep Learning Model For False News Classification and Identification

Abstract

The popularity of fake news has become a significant issue in contemporary society due to the rapid growth of internet social media and the ease of sharing information. The unintended consequences of spreading false information include social unrest, political polarization, and a loss of trust in media sources. The paper explores the performance of a deep learning method for automating the separation of fake and real news articles. The hybrid RNN +GRU model has been proposed and the classification performance of the model is highly reliable with an accuracy, precision, recall and F1-score equal to 99.8 on the test data. Further comparative analysis reveals that the proposed model performs greatly except the models that are currently being used, including CNN, Random Forest, XGBoost, ALBERT, and LSTM model. The findings confirm that the RNN + GRU architecture is an efficient and reliable method for false news detection. The paper provides a useful hybrid RNN+GRU model of the reliable fake news detection and it has high levels of robustness and stability as well as a high level of consistency in comparison to the current systems. In general, the results indicate that hybrid sequential models are indeed appropriate for detecting false news in real-world conditions.