Online Fake News Detection using Pre-trained Embeddings

Abstract

Fake news is a word, often heard. We often come across this term and its dangerous consequences with respect to society. To curb the menace of fake news is a big challenge in today's world. Many efforts have been made through different approaches to detect the fake news on online media. In this study, we propose a fake news detection system that utilizes pre-trained embeddings for the detection of fake news. We use a benchmark dataset of social media to carry our experiments. This study uses pre trained GloVe embedding based embedding matrix to generate the embeddings for the text as an input to our deep learning model. We also compare the performance of simple deep neural network based model with the LSTM based deep neural network on the same input embeddings. The results are indicative of good performance of the models to detect fake news.