TADM: A Trust-Aware and Drift-Adaptive Framework for Intelligent Data Management

Data management systems increasingly operate in dynamic environments where data distributions evolve continuously, and privacy risks are unpredictable. Most existing methods rely on static trust assumptions and threshold-based drift detection, leading to unstable governance decisions, delayed risk mitigation, and unnecessary utility loss. We propose a Trust-Aware and Drift-Adaptive Framework for Intelligent Data Management (TADM), which integrates a trust-aware and drift-adaptive data management framework that integrates dynamic trust modeling, soft drift awareness, and multi-objective optimization within a closed-loop control architecture. Trust is modeled as a time-dependent system state that jointly reflects data reliability, consistency under drift, and privacy risk, with drift severity incorporated directly into governance objectives through smooth, non-threshold-based penalties. TADM using both simulated data streams and a real-world dataset of NYC taxi trips. In synthetic experiments, TADM achieves stable governance, measured by fewer than one policy switch on average and a policy churn rate of 0.011, with fewer than one policy switch on average while preserving over 40% utility under moderate drift and degrading gracefully under severe drift. In the real-world case study, TADM achieves a mean trust level of 0.69, stable non-oscillatory policies, and over 63% utility, despite persistent non-stationarity, without requiring labeled drift events or manual intervention.