Case-Study-Based Review of Approaches for Transforming UML Class Diagrams to OWL and Vice Versa

Abstract

Building ontologies in The Web Ontology Language (OWL) as a knowledge representation about a particular domain gained a lot of interest over the recent years. Thanks to its large community and many options concerning tooling and methods for representation and transformations, OWL is being used not only in Linked Open Data and Artificial Intelligence but also in conceptual modelling. OWL allows capturing concepts and their properties, including relationships which can also be done using traditional conceptual models, for example, in Unified Modelling Language (UML). Both UML and OWL have their own specifics when compared to each other, and one may be more suitable than the other in concrete cases. There are several methods for transformation between knowledge representation in OWL and UML. In this paper, we review key methods for transforming UML to OWL and vice versa. To compare the methods, we use a non-trivial conceptual model that contains all commonly used constructs, e.g., generalization sets, composition, or relationships with constraints. The methods are evaluated in terms of information loss during transformation, the need for human intervention, and versatility.