Comparison of Fake News Detection using Machine Learning and Deep Learning Techniques

Fake news has spread widely on the Web in recent years due to the massive amount of information exchanged on digital media. This has motivates our study to determine the best-performing model among two Machine Learning models: Naïve Bayes (NB), Support Vector Machine (SVM), and three Deep Learning models: Long Short-Term Memory (LSTM), Neural Network with Keras (NN-Keras), and Neural Network with TensorFlow (NN-TF). We examined five models using two different English language news datasets. The performance of the models was evaluated using four metrics; accuracy, precision, recall and F1-score. The obtained results showed that deep learning models had achieved better accuracy than traditional ML models. The LSTM model has outperformed all other models examined. It achieved an average accuracy of 94.21%. The NN-Keras has also produced a good performance with an average accuracy of 92.99%. The words’ order carries critical information and plays a significant role in the fake news classification, where our LSTM makes a prediction based on this.