Mining hierarchical semantic periodic patterns from GPS-collected spatio-temporal trajectories

A large number of spatio-temporal trajectory data is being generated from GPS enabled devices such as cars, smartphones, and sensors. These trajectory datasets representing objects' movements provide new opportunities for enhanced spatio-temporal periodic pattern mining. These GPS collected trajectory datasets represent real-world movement phenomena and thus they are spatially placed, temporally recorded, aspatial semantically meaningful, hierarchically structured, and irregularly sampled. Periodic pattern mining from spatio-temporal trajectories is to find temporal regularities from these spatio-temporal trajectories, and thus must take into these five characterisics into account in order not to miss any spatio-temporally, semantically and hierarchically meaningful patterns from irregularly sampled spatio-temporal trajectories. Traditional periodic pattern mining fails to consider these five conditions simultaneously, and in this paper, we propose a hierarchical clustering based semantic periodic pattern mining to consider the five aspects: spatiality, temporality, semantics, hierarchy, and irregularity. Experimental results demonstrate the effectiveness of our proposed method against traditional periodic pattern mining approaches.

Expert Systems with Applications

Expert Systems with Applications

122

1873-6793

N/A

N/A

17