Uml-based design of visual languages for data warehousing

Abstract

Data warehouses are complex database systems for decision support, Their design has to take into account both the data sources that populate the data warehouse and the information requirements of the involved decision makers. Such data warehouse is indeed a database oriented to empower the performance of specific type of queries, in particular, those that manage data as multidimensional structures that can be shown from different viewpoints and granularities. On the other hand, UML has become the 'de facto' standard for software modelling. This unified language provides engineers with visual primitives and notions that facilitate the understanding of software systems by means of a compliant graph-based visualisation. This Ph.D. dissertation deals with the application, adaptation, and evaluation of UML to model several important layers of the data-warehousing architecture: data mining techniques, data-warehousing flows, information requirements, end-user metadata, among others.