The T-ADAPT site of Dr. Michael Hahne



Model Management

Model Transformation

SAP BW Models

SQL Server Models

Visio Shapes

Hahne online



Conceptual design of temporal aspects in data warehousing with T-ADAPT

from Dr. Michael Hahne

Chapter 1: Introduction

The increasing strategic adjustment of Information Technology with a stronger focusing on aspects of analysis and decision support is accompanied by the innovative technical concepts of Data Warehousing, OLAP and Data Mining. The possibilities of modelling determine decisively the efficiency and successful use of such systems.

Multidimensional views on enterprise-internal and -external data ensure useful approximations to the enterprise image of the manager and examine general, economically relevant circumstances considering numerous factors of influence. The revenue of a product group in certain sales regions during a defined period which is asked for is an example.

The arrangement of economical variables resp. indicators like revenue or gross sales is organized along different dimensions such as customers, articles and regions. This structuring is considered as a suitable decision-oriented view on economical facts. To put it figurative, the quantitative characteristics are stored in multidimensional cubes and their edges are marked and defined by individual dimensions [To03, 71; GCG00, 75].

There are two different types of dimensions. On the one hand there are small dimensions resulting from listing individual elements and on the other hand there are the dimensions in which elements are grouped in classes, the so called levels [PR03, 94; Gl01, 78; Ha02, 194]. These two types are called dimension defined by elements and dimension defined by levels. The constituent components of a dimension structure for dimensions defined by levels are shown in figure 1.

Fig. 1. Constituent components of dimensions

The time relation for data in multidimensional models is elementary, which is also expressed in the special meaning of time dimension because the facts and transaction data are inherently time stamped by this dimension [St01, 112]. The question about handling structural changes in dimensions, which arises again and again during the modelling process, is crucial for the structure of Data Warehouse and OLAP systems because thereby the possibilities of analyses are determined. The discussion of these temporal aspects in the Data Warehouse is presented in section 2.

The modelling of analysis-oriented information systems requires special care because thereby analysis possibilities are specified. Therefore suitable description possibilities are already needed in the conceptional phase. ADAPT (Application Design for Analytical Processing Technologies) which is described in section 3 is an acknowledged method for this. The linking of requirements of temporal aspects by modelling is discussed in section 4, in which ADAPT is extended by suitable temporal objects. Section 5 summarizes the results of the article and shows possible extensions.

Last update: 2010-01-13