Summarizability in Multiversion Data Warehouse

Abstract

Over time, data warehouse (DW) schema and instances may undergo several changes reflecting the changes that occur in data sources or the emergence of new user analysis requirements. To deal with these changes and ensure the persistence of historical states of DW, we propose a multiversion DW model that ensures the temporal versioning of DW schema components as well as of its instances. Our model supports heterogeneous multiversion dimensions which consist of optional levels having null values among their members. To ensure summarizability in such complex multidimensional structure, we propose to redefine the multiversion DW schema using temporal functional dependency with null values. Besides, we define specific dimension and fact constraints.