Content-based fake news detection using deep learning techniques: Analysis, challenges and possible solutions

Since the past decade, social media platforms are used widely for sharing information which may or may not be credible. Due to this many incidents have occurred that show how false news can have detrimental effect on the people and hence it becomes imperative to check the credibility of online news articles. Platforms like Facebook and Twitter have also taken multiple initiatives in past years to check the spread of misleading content on their platforms. Researchers have been continuously working on providing solutions to tackle the issue of fake news and its implications. This paper focuses on the Content-Based Fake News Detection (FND) which deploys text-based and visual-based approaches. In the past few years, a more sophisticated approach i.e the Multimodal approach is being used that combines the features of textual and visual data. This paper highlights various techniques for content-based FND using Deep Learning (DL) approach. These deep learning models and frameworks have been analyzed in detail and the results depict that there is a considerable improvement in the overall performance when the multimodal methods are used. Also, various techniques that might help for improving the efficiency of the FND frameworks are also included as future directions. This paper also presents numerous challenges that a researcher may encounter while modelling a FND framework and provides probable solution that might be applicable to overcome these challenges.