A model-driven approach for the multidimensional design of data warehouses

Abstract

Designing a multidimensional model of a data warehouse is a highly complex, prone to fail, and time consuming task, due to the fact that (i) the information needs of decision makers and the available operational data sources that will populate the data warehouse must be both considered in a conceptual multidimensional model, and (ii) summarizability-aware non-trivial mappings must be performed to obtain the final implementation of this conceptual multidimensional model, However, no significant effort has been made to take these issues into account in a systematic, well structured and comprehensive development process. To overcome the lack of such a process, this PhD Thesis proposes a model-driven approach for the development of a hybrid multidimensional model at the conceptual level and for the automatic derivation of its logical representation as a basis of the implementation. A normalization process is also proposed to avoid summarizability problems at the conceptual level and in further implementation stages. Finally, an Eclipse-based tool has been implemented as a proof of concept of this research. A case study has been implemented by using that tool, which shows each step of the presented approach.