PL-NCC: A Novel Approach for Fake News Detection through Data Augmentation

Abstract Fake news has become rampant with the prominence of social media, jeopardizing information integrity and the reliability of news. It becomes imperative for media platforms to develop robust and efficient strategies to mitigate the spread of fake news. We make two contributions to fake news research in this work. Firstly, we propose the Psycho-Linguistic News Content and Comments (PL-NCC) dataset, a consolidated dataset of two leading fake news datasets, NELA-GT and Faked-dit, which leverages linguistic and psychological characteristics from the news articles and user comments to improve the classification accuracy of benchmark models. Secondly, we propose the News Content and Comments (NCC) classification model to leverage the psychological features extracted from our PL-NCC dataset, which introduces a feed-forward layer to a deep learning model, enhancing the efficacy of the extracted psychological features to better identify fake news. Our approach achieves a classification accuracy of over ninety percent, exceeding baseline results.