Towards an Ontology of Multidimensional Data Structures for Analytical Purposes

Abstract

Multidimensional data are the foundation for OLAP applications. They can be provided in several ways: relational OLAP, multidimensional OLAP, or hybrid OLAP. The usage of the underlying technology, which is well understood and in most cases formally defined, does not resolve the issue of a missing vocabulary for multidimensional data on a conceptual level. Some basic definitions are broadly used; for example cube, dimension, and the operations like slice, dice, roll-up, and drill-down. When it comes to more sophisticated constructs like irregular hierarchies, different vocabulary exists.For the integration of different OLAP applications as well as for an easier development (each stakeholder has to use the same vocabulary in order to reduce misunderstandings and projects failures), we provide an approach for a comprehensive ontology of multidimensional data. It defines the vocabulary used during the design of OLAP and data warehouse applications. The ontology can be seen as a basis for (i) notation assessment by evaluating the notation against the BungeWandWeber model, (ii) the ontological engineering of new data warehouse and OLAP applications, and (iii) ontological model integration.