A Multichannel CNN-LSTM Approach for Fake News Detection: A Comparative Study

The proliferation of fake news across various language sources poses a significant challenge in ensuring the dissemination of accurate and reliable information. Addressing this issue is crucial for maintaining the integrity of news content and public trust. We developed a novel approach combining Convolutional Neural Networks (CNN) for deep feature extraction and Long Short-Term Memory (LSTM) networks for subsequent detection of fake news. Our approach utilizes a Multichannel combined CNN-LSTM architecture. We collected an extensive dataset comprising 40,000 unique news articles from diverse sources to evaluate our model. Additionally, we integrated Gaussian and Random Forest models to benchmark performance. Our findings revealed that the LSTM component of the model achieved a remarkable accuracy of 97%, significantly outperforming the CNN's accuracy of 75%. The Gaussian model also recorded an accuracy of 75%, while the Random Forest model matched the LSTM's performance with an accuracy of 97%. The study demonstrates the effectiveness of our hybrid CNN-LSTM approach in detecting fake news, highlighting its potential as a robust tool for combating misinformation. Our results contribute to the growing body of research in fake news detection and underscore the value of leveraging deep learning techniques to enhance the accuracy and reliability of news content.