Abstract
Business Intelligence is often described as a set of techniques serving the transformation of raw data into meaningful information for business analysis purposes. Thanks to the technology development in the realm of Geographical Information Systems, the so-called trajectory data were appeared. Analysing these raw trajectory data coming from the movements of mobile objects requires their transformation into decisional data. Usually, the Extraction-Transformation-Loading (ETL) process ensures this task. However, it seems inadequate to support trajectory data. Integrating the trajectory aspects gives the birth of Trajectory ETL process (T-ETL). Unfortunately, this is not enough. In fact, the business analysis main purpose is to minimize costs and time consuming. Thus, we propose to swap the T-ETL tasks scheduling: instead of transforming the data before they are written, the Trajectory Extraction, Loading and Transformation (T-ELT) process leverages the target system to achieve the transformation task. In this paper, we rely on a set of powerful mechanisms to handle the complexity of each T-ELT task. Wherefore, an algorithm is dedicated to ensure the transformation of raw mobile object positions into trajectories and from there we highlight the power of the Model-driven Architecture approach to transform the resulting trajectories into analytical data in order to perform the Business Intelligence goal.