Model-Based Requirements Engineering for Data Warehouses: From Multidimensional Modelling to KPI Monitoring

A Data Warehouse (DW) is one of the main components of every BI system. It has been convincingly argued that the success of BI projects can be strongly affected by the Requirements Engineering (RE) phase, when the requirements of a DW are captured. Multiple RE methods for DWs have been proposed which have goal models in the core of their approach. Existing methods cover RE up to the static part of a DW, where the Multidimensional (MD) model is obtained. However, the RE for the dynamic part of the DW, where the requirements of operations on the DW are captured, has been neglected in the literature. In this paper, we propose a RE method, covering both the static and the dynamic part of a DW in an integrated manner. Our approach is to use the concept of a Key Performance Indicator (KPI). We initially use KPIs as the main driver to obtain the MD model and then discuss how decision-makers analyse them in order to measure the success of an organisation. In our method, the goal model from the i* framework was extended with UML use case diagrams.