Positionalism of Relations and Its Consequences for Fact-Oriented Modelling

Abstract

Abstract. Natural language-based conceptual modelling as well as the use of diagrams have been essential components of fact-oriented modelling from its inception. However, transforming natural language to its corresponding object-role modelling diagram, and vv., is not trivial. This is due to the more fundamental problem of the different underlying ontological commitments concerning positionalism of the fact types. The natural language-based approach adheres to the standard view whereas the diagram-based approach has a positionalist commitment, which is, from an ontological perspective, incompatible with the former. This hinders seamless transition between the two approaches and affects interoperability with other conceptual modelling languages. One can adopt either the limited standard view or the positionalist commitment with fact types that may not be easily verbalisable but which facilitates data integration and reusability of conceptual models with ontological foundations.

Toward an Ontology-Driven Unifying Metamodel for UML Class Diagrams, EER, and ORM2

“…The second difference between relationship and object type is due to roles, which are called "association ends" or "member ends" in UML [31] "roles" in ORM and fact-based modelling [19,20,11], components of a relationship in EER [10,35]. A role is something that an object plays in a relationship, and, thus, a relationship is composed of at least two roles, therewith contributing to the characterisation of relationships and committing to the so-called positionalist ontological commitment of relations and relationships (see [28] for a good overview, which has been applied to ORM in [25]). The consequences are that all three CDM languages have roles-hence, logically, they have to be first-class citizens in a formalisation-and, ontologically, this inclusion entails that from an information modelling viewpoint, they form a part of the so-called 'fundamental furniture of the universe' and thus that they are ontologically distinct from entity types and relationships.…”

“…The differences between the main conceptual data modelling (CDM) languages-UML Class Diagrams, ER, EER, ORM, and ORM2-may seem merely terminological, but it is known that from a metamodelling viewpoint, this is not the case [18], and at times not even within the same family of languages [25]; conversely, what may seem different may actually not be, or at least have a common 'parent' in meaning. The latter concerns differences in ontological foundations, but the state of the art in this area has not gone beyond a single CDM language and only for UML and ORM (e.g., [15,25]).…”

“…The latter concerns differences in ontological foundations, but the state of the art in this area has not gone beyond a single CDM language and only for UML and ORM (e.g., [15,25]). Thus, it is unclear to what extent the languages differ or agree on their underlying principles for modelling information.…”

Structural Entities of an Ontology-Driven Unifying Metamodel for UML, EER, and ORM2

“…However, this capability is not common in traditional information systems development and management other than at the physical schema layer [6] and for conceptual models represented in the same language [2,12]. Subtle representational and expressive differences (e.g., [17]) in the languages are primarily due to their different origins and purposes, and makes this task very difficult, and even within one language family there are differences in meaning of an element [14,24].…”

“…The state of the art in this area has only incidentally gone beyond a single Conceptual Data Modelling (CDM) language and only for UML and ORM (e.g., [14,24,25]). It is unclear to what extent the languages agree on their underlying ontological foundations to model information.…”