Abstract
Since the introduction of the multidimensional data model as modeling formalism for Data 
Warehouses (DWs) design, several techniques have been proposed to capture multidimensional 
properties at the conceptual level. 
These techniques have mostly focused on different specific fields such as: user requirements, 
operational database schema analysis or mixed approaches (is not only initialized from user’s 
requirements, but also from the operational database schema).
On the other hand, MDA is a new standard to develop software by transforming models, for this 
MDA proposes three viewpoints: Computation Independent Model (CIM), Platform Independent 
Model (PIM) and Platform Specific Model (PSM). 
This thesis is developed in the context of conceptual design of DWs with MDA (a methodology 
for DW conceptual design). This method is employed within a mixed approach and consists of 
three phases.
The first phase, is devoted to examining the ER schema of the operational database, finding out 
candidate multidimensional schemas for the DW, this is achieved through a set of 
transformation rules between the Entity Relationship (ER) PIM and the On-Line Analytical 
Processing (OLAP) PIM.
In the second phase, user requirements are collected. The purpose of this phase is to gather 
information from business analysts and/or managers about the company goals and needs. For 
the basis of our elicitation method, we adopt a goal model. 
The final phase integrates these two viewpoints, and thus generates a feasible solution 
(supported by the existing data sources) that best reflects the DW user’s requirements.