Title: A shallow-based neural network model for fake news detection in social networks
Authors: S.P. Ramya; R. Eswari
Addresses: Department of Computer Applications, National Institute of Technology Tiruchirappalli, India ' Department of Computer Applications, National Institute of Technology Tiruchirappalli, India
Abstract: The convenience of connecting through the internet and eagerness to spread any news through online social media is very intriguing as it can be done rapidly and with very little effort. This permits the very quick spread of fake news globally and misleads the people against democracy and freedom. The content of fake news very closely resembles true news. So, technically, it is tough for a deep neural network to 'detect and attend to' the 'fake only' aspects of a news article. Fake news detection is a significantly complex and challenging task from the aspect of deep learning-based attention mechanisms. The deep learning-based fake news detection systems suffer from indistinguishability of fake and real news/data, the curse of high dimensionality, the high training time of deep neural networks, the over-fitting of the network training, and the over-thinking problem. In this paper, a shallow-based convolution neural networks (SCNN) model has been proposed for the fake news detection system to overcome the mentioned issues. The proposed SCNN model is experimentally tested for a complex benchmark LIAR dataset. The performance of the proposed SCNN is better than other existing models in terms of accuracy, precision, recall and F1-score.
Keywords: attention mechanism; deep learning; optimisation; natural language processing; NLP; convolution neural networks; CNN.
DOI: 10.1504/IJICS.2023.132727
International Journal of Information and Computer Security, 2023 Vol.21 No.3/4, pp.360 - 382
Received: 11 Sep 2021
Accepted: 16 Dec 2021
Published online: 09 Aug 2023 *