Simulating Our LifeSteps by Example

During the past few decades, a number of effective methods for indexing, query processing, and knowledge discovery in moving object databases have been proposed. An interesting research direction that has recently emerged handles semantics of movement instead of raw spatio-temporal data. Semantic annotations, such as “stop,” “move,” “at home,” “shopping,” “driving,” and so on, are either declared by the users (e.g., through social network apps) or automatically inferred by some annotation method and are typically presented as textual counterparts along with spatial and temporal information of raw trajectories. It is natural to argue that such “spatio-temporal-textual” sequences, called semantic trajectories, form a realistic representation model of the complex everyday life (hence, mobility) of individuals. Towards handling semantic trajectories of moving objects in Semantic Mobility Databases, the lack of real datasets leads to the need to design realistic simulators. In the context of the above discussion, the goal of this work is to realistically simulate the mobility life of a large-scale population of moving objects in an urban environment. Two simulator variations are presented: the core Hermoupolis simulator is parametric driven (i.e., user-defined parameters tune every movement aspect), whereas the expansion of the former, called Hermoupolisby-example, follows the generate-by-example paradigm and is self-tuned by looking inside a real small (sample) dataset. We stress test our proposal and demonstrate its novel characteristics with respect to related work.