An ontological model of information systems, the Bunge-Wand-Weber (BWW) model, is used to analyse and evaluate the Unified Modeling Language (UML) as a language for representing concrete problem domains. As a result, each relevant and major UML construct becomes more precisely defined in terms of the phenomena in and aspects of the problem domain it represents. The analysis and evaluation shows that many of the central UML constructs are well matched with the BWW-model, but also suggests several concrete improvements to the UML-metamodel. New metaclasses are proposed to distinguish between (physically) impossible and (humanly) disallowed events, based on UML-exceptions. New abstract metaclasses are proposed for static and behavioural constraints, behaviours and static behaviours, as well as binding relationships and coupled events. New meta-subclasses of UML-objects, -classes, -types and -relationships are proposed to make the UML more orthogonal, and a new definition is proposed for UML-responsibilities. The analysis also shows that the constructs in the UML must play several roles simultaneously, supporting representation both of the problem domain, of the development artifacts and of the proposed software or information system, while fitting together as a tightly integrated, well-defined language.Keywords: Object-oriented analysis -Problem domain representation -Ontological analysis and evaluationUnified Modeling Language (UML) -The Bunge-WandWeber model (BWW)
BackgroundThe Unified Modelling Language (UML) [22] has become the de facto standard for object-oriented (OO) * Corresponding author modelling during information systems (IS) development. However, the suitability of the UML for modelling concrete problem domains in the early development phases has been called into question [25]. The suitability of object-oriented modelling in general in the early development phases is also controversial [17,33]. At the same time, many authors [7, 28] have argued that the early development phases are critical for successful and cost efficient IS development. In consequence, an OO language like the UML, if not tightly integrated and well defined, might exacerbate rather than ameliorate current problems in the IS development industry.This paper analyses and evaluates the major constructs of the metamodel of the stable Version 1.3 of the UML [22] in terms of how well the modelling constructs it provides are suited for representing concrete problem domains. The analysis and evaluation is anchored in the Bunge-Wand-Weber (BWW) model of information systems. The BWW-model, e.g., [37,38,41], is an adaptation of Mario Bunge's comprehensive ontology [3,4], which is inspired by systems theory. Weber [42] points to ontology as the branch of philosophy that deals with theories about the nature of things in general as opposed to theories about particular things. Ontological theory is therefore well-suited for benchmarking the adequacy and sufficiency etc. of modelling constructs for representing concrete problem domains. This paper i...