Published Paper Details:

Modern digital platforms use AI to show users personalized ads and recommendations based on their past behavior. However, most users do not clearly understand why a particular product or advertisement is shown to them. This lack of transparency often creates confusion and reduces trust in such systems. In this paper, we present T-Trace, a simple and transparent framework that helps track how a user's past actions influence their current decisions. Instead of only explaining why an ad appears, the system tries to connect a user's final decision (such as making a purchase) with earlier activities like searching, browsing, or comparing products. The proposed system collects user interaction data and processes it using a lightweight machine learning model. It then uses SHAP-based explanations to highlight which past actions had the most impact on a particular decision. To maintain reliability, each decision and its explanation are stored in a hash-linked structure, making it difficult to alter past records. We also developed a Streamlit-based dashboard that presents these insights in an easy-to-understand format using timelines and simple explanations. Through this approach, the system aims to improve transparency, help users better understand their own decisions, and support ethical use of AI in personalization systems.