Development and evaluation of an ontologically complete and clear object-oriented conceptual modelling grammar

Conceptual modelling is an Information System (IS) modelling genre that is carried out during the analysis phase of the IS development process. It is used to build a representation of the types of phenomena that exist in IS users’ real-world scenarios. The conceptual models produced are then used as the basis for design in subsequent phases of the IS development process. The Object-Oriented (OO) methodology is a widely used IS development methodology. Nonetheless, several research studies have concluded that modelling grammars that support the OO methodology focus more on modelling system design and implementation phenomena than the real-world phenomena in IS users’ domains. Thus, OO grammars have limitations as a conceptual-modelling tool. Such limitations may result in defective conceptual models, which could then undermine the quality of the IS that ultimately is developed. The purpose of this research study was to enhance OO modelling grammars to make them more appropriate for conceptual modelling. The research work focused on one widely used OO modelling grammar—namely, the Unified Modelling Language (UML). Enhancing the UML grammar to make it more appropriate for conceptual modelling was carried out in two phases. In the first phase, a novel UML grammar was developed through modifying the existing UML grammar using the following steps: (1) Evaluating the constructs in all 14 UML diagrammatic notations. The intent was to identify a subset of UML constructs that are capable of representing real-world phenomena in user domains. (2) Evaluating the completeness and clarity of this subset of UML constructs. Completeness means the UML constructs are able to model any type of real-world phenomena that interests a user, while clarity means the grammatical constructs represent each and every type of real-world phenomenon uniquely. (3) Removing the deficiencies identified in the UML grammar by adding some new grammatical constructs and modifying or removing some existing grammatical constructs. (4) Evaluating the UML diagrammatic notations to identify those that can and those that cannot be used for conceptual modelling. The second phase of the research study was an empirical evaluation of the novel UML grammar. By undertaking the empirical study, an industry perspective has been added to the theoretical perspective used in the first phase. The evaluation procedure was carried out in two cycles with two IS development projects undertaken by a Sri Lankan software development organization. The conceptual modellers who participated in the two IS projects were trained in the novel UML grammar. They undertook conceptual modelling using the novel UML grammar. In each cycle, feedback on the novel UML grammar was obtained not only from the modellers but also other project team members. The results of this empirical evaluation suggest that the novel UML grammar facilitates high-quality conceptual modelling in practice. Although the research has focused on UML, the approach developed can be applied to any OO modelling grammar. The outcome should be OO modelling grammars that are better suited to conceptual-modelling tasks, thereby allowing practitioners to produce higher-quality conceptual models and ultimately higher-quality IS designs.