A density-based approach for mining movement patterns from semantic trajectories

In this paper, we study the problem of discovering all movement patterns from semantic trajectory databases. We propose a two-step method to solve this problem efficiently. We first retrieve frequent movement patterns of categories from the transformed database of sequential categories, and then cluster dense trajectories in a growth-type way for all movement patterns. Moreover, we define a new metric distance function on trajectories. We also use M-tree to cluster trajectories more efficiently. Our experimental results demonstrate the efficiency of the proposed method.