Generating contextual trajectories from user profiles

The trajectory of traffic participants is an essential source for pattern mining and knowledge discovery in urban mobility. However, real-world trajectory data are often not publicly available due to privacy concerns or intellectual property constraints. Although some simulators or synthetic trajectory datasets have been proposed, many of them only consider the spatial-temporal aspects of the trajectory data, but ignore other contextual information that could impact trajectories. On one hand, trajectories are usually associated with and affected by user profiles (e.g., a person's daily routines and preferred modes of transportation). On the other hand, an individual's movements are also affected by environmental conditions and interactions with other traffic participants, particularly in urban scenarios (e.g., routing choices due to congestion or road conditions). Such contextual trajectories provide a more realistic representation of the mobility patterns of traffic participants. Due to the lack of such datasets or trace generators, this work presents ConTraSim (<u>Con</u>textual <u>Tra</u>jectory <u>Sim</u>ulation), a novel approach for generating contextual trajectories based on the Simulation of Urban Mobility (SUMO) traffic simulator. More specifically, the proposed approach is designed to produce GPS traces annotated by contextual information that mimic the movements of multiple types of traffic participants in urban areas. As a case study, we also generate a sample dataset using the proposed method and compare it to real-world data to demonstrate how well the synthetic data reflects real-world data characteristics.