Transportation Mode Annotation of Tourist GPS Trajectories Under Environmental Constraints

Abstract

Tourist transportation usage analysis provides basic information for tourism policy making. With the technical advances of tracking devices, GPS-equipped smartphones sense the movement of tourists and generate extensive volumes of movement data detailing tourist trajectories. Many researchers study semantic annotation using machine learning. However, it is necessary for machine learning to label the data for training; this requirement is costly. It would be useful for GPS semantic annotation if labelling the substantial amounts of GPS data could be avoided. In this research, we propose a new, simple GPS semantic annotation method using environmental constraints without machine learning. We call this method Segment Expansion with Environmental Constraints (SEEC) and assume a tourist behaviour model in which tourists move by foot and public transportation in touristic destinations that include numerous locations of interest. SEEC inferred the transportation modes of the GPS trajectory data at a 90.4 % accuracy level in the experiment.