Computer-Aided Data-Mart Design

Abstract

With decision support systems, decision-makers analyse data in data marts extracted from production bases. The data-mart schema design is generally performed by expert designers (administrator or computer specialist). With data-driven, requirement-driven or hybrid-driven approaches, this designer builds a data-mart defining facts (analysis subjects) and analysis axes. This process, based on data sources and decision-makers requirements, often turns out to be approximate and complex. We propose to design a data-mart schema by the decision-maker himself, following a hybrid-driven approach. Using an assistance process that visualises successively intermediate schemas built from data sources, the decision-maker gradually builds his multidimensional schema. He determines measures to be analysed, dimensions hierarchies within dimensions. A CASE tool based on this concept has been developed.