Evolution of a Dynamic Multidimensional Denormalization Meta Model Using Object Role Modeling

Abstract

At Guidant, a Boston Scientific Company, systems that collect data in support of medical device clinical research trials must be capable of collecting large, dynamic sets of attributes that are often reused in later research activities. Their resultant design, based on conceptual analysis using Object Role Modeling (ORM), transforms each unique business fact into an instance in a highly normalized star schema structure with related dimensions. When it becomes necessary to generate focused denormalized reporting structures from this star schema, hereafter referred to as miniature data marts or simply “mini marts”, the dynamic nature of these source attributes can present a maintenance challenge. Using ORM, we propose a meta model that supports the definition, creation, and population of these denormalized reporting structures sourced from a multidimensional fact table that also leverages a hierarchical taxonomic classification of the subject domain.

An erratum to this chapter can be found at http://dx.doi.org/10.1007/11915072_109.