Explainable AI-Based Detection of Misinformation Spread in Online Social Networks

Abstract

Social media is rapidly expanding and has completely transformed how people exchange and receive information. Social media platforms like Facebook, Instagram, and Twitter allows us to connect with others across the globe instantly and make the communication very easy and fast than ever before. Social Media platform is very convenient to share the useful information as well as fake information is spreading swiftly. In case the information is not true, people get influenced by the public opinion and act accordingly. and even lead to serious social or political consequences. As a result, identifying this false information in huge amounts of social media data has emerged as a significant research topic. Machine learning and artificial intelligence have found widespread application in the automated identification of misinformation. Nevertheless, many existing models function as "black-box" systems, thereby obscuring the processes underlying their decision-making When automated systems don’t clearly show how they make decisions, people may start to question how reliable they really are. To find this issue, this research suggests using Explainable AI (XAI) in misinformation detection tools. By making the decision-making process more transparent and easier to understand, XAI can help build trust and make these systems feel more dependable. This framework clarifies the prediction generation process, utilizing machine learning and natural language processing methods, and employing tools like SHAP and LIME. These indicate that Explainable AI enhances the performance and transparency of misinformation detection. As a result, this methodology can bolster the dependability of detection systems, thereby fostering a more thorough comprehension of false information dissemination for researchers, social media platforms, and policymakers.