From Conceptual Modeling to Logical Representation of Trajectories in DBMS-OR and DW Systems

Abstract

With the growth of mobile devices equipped with geographical localization services, it has become economical and technically feasible to capture moving object trajectories in real life. Many interesting applications are being developed based on trajectory analysis. For example, in a traffic management system, traffic jams may be determined by mining movement patterns of groups of cars. In general, trajectory data can be analyzed into two different perspectives: real time and historical. In addition, trajectory applications share a common need, a more structured recording of movement. This allows managing trajectories as first class citizens and attaching whatever semantics the application requires, and developing robust and efficient methods to aggregate a set of trajectories that may support complex analysis. This paper extends previous work on the conceptual modeling of trajectories by generalizing the idea of stops and moves and by defining a set of aggregation functions on trajectory data. In addition, this work proposes two modeling approaches, both based on design patterns, for devising trajectory data schemas for relational and multidimensional environments. A real-world case study about truck trajectories is used as a proof of concept. Experimental results showed that these modeling approaches offer the flexibility we are looking for to cope with the potential complexity of trajectory semantics in real time and historical analyses.