An ontological comparison and evaluation of data modelling frameworks

Data modelling frameworks are used to construct models of reality for use in information systems, computer science and software engineering. Currently, there is a plethora of data modelling frameworks each possessing a view of the world, in that each data modelling framework has a distinct set of terms that are used to create models of reality. We are interested in finding a unifying framework in which to compare and study data modelling frameworks. We turn to the study of ontologies to find a theory with potential as a unifying framework. Ontologies are studied in philosophy and are concerned with 'what there is'. An ontology defines the categorial structure of reality and the terms that are fundamental for describing reality. Consequently, ontologies are ideally suited to comparing and evaluating data modelling frameworks, and an ontology has the potential to provide a unifying framework for data modelling frameworks. In this thesis we consider using ontology in this role and examine using a specific commonsense realistic ontology from philosophy in the role of a unifying framework. We are seeking to establish the nature and degree of synonymity between the world view of a pragmatically selected ontology and of several representative data modelling frameworks. We propose and apply two qualitative methods to help us. By applying these methods, we can begin to understand the efficacy of our approach and the generality of using an ontology as a unifying framework. We have found that there is reasonable commonality between the world view of the selected ontology and that of the data modelling frameworks we have studied. However, we have found areas in which specific data modelling frameworks do not match the world view of the ontology. We have also found areas in which aspects of specific data modelling frameworks are not supported by the selected ontology. Most notably, we have found grounds to question the rigid class hierarchies common in many object modelling frameworks. However, all data modelling frameworks appear capable of being extended to support the selected ontology. We have reason to believe that the selected ontology is thus an excellent candidate for further investigation as a unifying framework.