XAI for User Classifications

Explainable artificial inte lligence (XAI) models are artificial intelligence (AI) systems comprising learned algorithms and outputs that users can trust and understand. In this study, an XAI method is suggested in the context of topic modeling, which allows users to understand how an AI system makes classifications. This method is an innovative procedure to re-engineer the process of user classification on the server side, applying mathematical calculations done on the user side. Our proposed method yielded highly accurate results. This method provides users with an improved understanding and awareness of how the system classifies their preferences. Therefore, users are able to control and adjust the recommendations they receive from the AI system.