An analysis and characterisation of publicly available conceptual models

Multiple conceptual data modelling languages exist, with newer version typically having more features to model the universe of discourse more precisely. The question arises, however, to what extent those features are actually used in extant models, and whether characteristic profiles can be discerned. We quantitatively evaluated this with a set of 105 UML Class Diagrams, ER and EER models, and ORM and ORM2 diagrams. When more features are available, they are used, but few times. Only 64\% of the entities are the kind of entities that appear in all three language families. Different profiles are identified that characterise how a typical UML, (E)ER and ORM diagram looks like