A New Data Warehouse Approach Using Graph

Abstract

The standard source data-based approach for data warehouse design, has always been criticized since it ignores the user's needs. However, this inefficiency is due to the fact that works adopting this trend, consider that the conceptual model or a physical data source is the own and best representative of data sources. All requests subject to operational production bases, which represent punctual needs, were hence always ignored. In this article we consider the history of these requests, in addition to the conceptual data model (CDM), in order to introduce a new approach for an automatic design of a multi-dimensional diagram. The CDM is first translated into a multivariate pattern. Then, it is refined by using the history of requests. Both design steps use graphs as a formalism to represent the data model decision-making.