Indoor mobility data encoding with TSTM-in: A topological-semantic trajectory model

Abstract

The growing ubiquity of location/activity sensing technologies has created unprecedented opportunities for research on human spatiotemporal interaction behavior in mobile environments. However, existing studies of human mobility need to sufficiently account for the association of indoor scenes with the semantics of human behavior. This paper introduces TSTM-in, a trajectory model that combines trajectory data and indoor scenes using topological semantic modeling, semantic trajectory reconstruction, and trajectory queries. The model effectively manages indoor semantic trajectory data and extracts topological behavioral semantics by incorporating important points across a trajectory to reflect the semantics of key points connected to indoor corridors and regions. These topological semantics facilitate the creation of a flexible intersection-based indoor semantic trajectory reconstruction. Reconstructed semantic trajectories represent human mobility by integrating semantic data sets along the time axis. A case study with real-world trajectory queries from travelers demonstrates the model's effectiveness. TSTM-in realizes the association of indoor scenes with human behavior semantics, supporting the construction of mobile object management applications for indoor scenes and providing scientific and reasonable spatiotemporal semantic information description for location service-based intelligent cities.