Trajectory-Based Spatiotemporal Entity Linking

Trajectory-based spatiotemporal entity linking is to match the same moving object in different datasets based on their movement traces. It a fundamental step support data integration and analysis. In this paper, we study problem of using effective concise signatures extracted from trajectories. This formalized as $k$ -nearest neighbor ( -NN) query signatures. Four representation strategies (sequential, temporal, spatial, spatiotemporal) two quantitative criteria (commonality unicity) are investigated for signature construction. A simple yet dimension reduction strategy developed together with novel indexing structure called WR-tree speed up search. number optimization methods proposed improve accuracy robustness linking. Our extensive experiments real-world verify superiority our approach over state-of-the-art solutions terms both efficiency.