MOA: A hybride multidimensional data model for interactive data exploration based on object-relational analysis units

Our approach to model and represent multidimensional data for explorative data analysis [16] is guided by our experience in the context of epidemiological cancer research. The project CARLOS (Cancer Registry Lower-Saxony) developed the Epidemiological and Statistical Data Exploration System (CARESS) to support multidimensional analysis of health data. The system is based on an architecture that focuses on extensive interoperability between a database management system and several analysis and visualisation tools. To meet the requirements we developed the concepts of static data spaces and dynamic analysis units as abstract data types, so-called MOA (Model for OLAP Applications). We integrated MOA into an extendible, object relational, database system to provide a platform for the developement of advanced support for