Fake News Detection Using Convolutional Neural Networks

Abstract

There exist catchy headlines in the digital media so as to lure readers to click on them. These exaggerate the facts and baits the users since the actual contents deviate farther from the clickbait headlines. A Machine learning model has been implemented using Convolutional Neural Network (CNN) to train and test on English dataset and then to perform classification. The CNN model was trained using Keras classifier with Tensor flow as backend. The model was trained and validated with the training dataset and validation dataset. Finally, the model is evaluated with the testing dataset. The CNN model obtained an F1 score of 92 % on test data. Cross validation technique was used for data validation.