A Layered Multidimensional Model of Complex Objects

Abstract

Multidimensional modeling is nowadays recognized to best reflect the decision makers’ analytical view on data. In this paper, we address some modeling features that we believe existing multidimensional models do not fully cover, such as considering real life entities that are meant to be analyzed as complex objects, allowing for simple and complex measures, treating facts and dimension members equally and observing hierarchies within and between complex entities. We propose a layered multidimensional model based on the concept of complex object which encapsulates data and structure complexity and eases the creation and manipulation of complex data cubes. We need to define our model at three layers. The first layer class diagram describes complex objects and captures the hierarchical organization of their attributes. The second layer package of classes describes the multidimensional model as a set of complex objects that are connected by relationships and some of which are organized in hierarchies. The third layer package of packages describes complex cubes which are derived from the multidimensional model. We show the benefits and feasibility of our proposals through their implementation in a real-life case study.Keywordscomplex objectcomplex relationshiphierarchymultidimensional modelcubeprojectionlayer