Datawarehouses et systèmes multidimensionnels : comparaison des méthodes de modélisation
Résumé
Many papers were dedicated to database conceptual design. This paper is concerned with conceptual design of datawarehouses and multidimensional systems. We describe and compare three approaches: - the method proposed by Kimball based on the star model, - the method proposed by Kimball based on the flake model, - the MAP method proposed by Akoka and Prat based on the multidimensional model. The three approaches are illustrated through the same example consisting of a decisional system which aims at evaluating companies for acquisition. Based on EER conceptual modeling, the last method allows the designer to distinguish strictly between conceptual and logical designs. Kimball's methods offer a rapid way to generate the metadata of the datawarehouse. Further research is discussed.