Trajectory-user linking identifies the likely user behind an anonymous mobility trace, but existing methods often lose fine-grained spatial, temporal, and semantic information or scale poorly. TRoPE-TUL develops a Transformer-based approach that better represents distances, time gaps, and nearby Points of Interest while handling sparse and irregular trajectories. The goal is to improve linking accuracy at lower computational cost, particularly under realistic class imbalance and heterogeneous mobility patterns. Experiments on real-world mobility datasets show strong performance against recent trajectory-linking methods.
Team: Ezharuddin Jubaer, Sumaiya Karim Katha, Mir Sazzat Hossain, Fahim Shahriar, Amin Ahsan Ali, AKM Mahbubur Rahman
