The Reveal of Fake News on Twitter Using Credibility Analysis and Multimodal Approach

Abstract

Misleading news is becoming a big threat to news accuracy since it is so easy to communicate and spread across various social media platforms. Fake news has a tremendous impact on society as a result of the proliferation of online social media. Twitter is one of the social media platforms that is regularly used to propagate misleading information during political campaigns. In this study, we propose a system for detecting bogus news that incorporates text and image analysis techniques, as well as credibility analysis. The system is divided into three modules: text processing, image processing, and credibility analysis. We pre-process textual data in the text processing module using the word embedding technique, which allows us to extract significant features and identify potential indicators of fake news. We focus on convolutional neural networks (CNN) to extract useful characteristics from images. The credibility analysis module examines Twitter data, gathering real-time tweets and employing various analysis approaches such as hashtag analysis, user analysis, and retweet analysis. We assess the credibility of the tweets by considering these aspects, providing additional insights into the overall credibility of the news events. The system has a 97% accuracy rate in categorising news item as fake or true.