Semantic Trajectory Clustering via Improved Label Propagation With Core Structure

Abstract

Traditional trajectory clustering algorithms mostly cluster trajectories from the perspective of temporal or spatial. However, trajectories with temporal and spatial similarity may be semantically related, and ignoring semantic information may lead to unreasonable trajectory clustering results. Besides, most of the existing semantic trajectory clustering algorithms cannot deal with overlapping trajectories well. In this paper, we propose a semantic trajectory clustering based on improved label propagation with core structure, which could better measure the semantic trajectories similarity and cluster semantic trajectories with overlapping spaces from the perspective of the network. Experimental results demonstrate that the proposed algorithm can effectively cluster semantic trajectory datasets compared with some traditional and recently proposed methods.