Article Preview
Top1. Introduction
Domain-specific modeling languages (DSMLs) are developed for creating models within specific domains by means of a strongly cohesive set of domain concepts (Henderson-Sellers, 2012). On the contrary, general-purpose modeling languages (GPML) consist of domain-independent concepts (e.g. UML, EER or BPMN). As a result, DSMLs enable the rapid modeling of the behavior and/or structure of applications in well-defined domains (Sprinkle & Karsai, 2004). Different types of DSMLs have been proposed. Executable DSMLs allow the creation of domain models that can be transformed into executable code. Visual DSMLs on the other hand describe aspects of the physical and social world for purposes of human understanding and communication (Mernik, Heering, & Sloane, 2005). These languages have been developed, for instance, to model different aspects related to economic reality such as the Architecture for Integrated Information Systems (ARIS) framework (Scheer, 1998) and value creation processes (Gailly & Poels, 2007a). In this paper, we will focus on visual DSMLs and henceforth refer to them as DSML.
In order to be effective, a DSML should be sufficiently expressive to represent the domain concepts that are captured by the intended models. To better fulfill these requirements, ontologies have been introduced as a theoretical foundation (Wand, Monarchi, Parsons, & Woo, 1995). For keeping a broad interpretation, we adopt the characterization of ontologies as described by Honderich (2006), which defines ontology as “the set of things whose existence is acknowledged by a particular theory or system of thought”. Ontologies support the construction of explicit models of conceptualizations in the form of concrete guidelines for selecting which concepts should be represented as language constructs and how they should be applied (Guizzardi, Pires, & Sinderen, 2002). Moreover, ontologies can be applied to evaluate the quality of a modeling language and its ability to describe a certain domain by performing an ontological analysis. An ontological analysis improves a DSML by: (i) providing a rigorous definition of the constructs of a modeling language in terms of real-world semantics, (ii) identifying inappropriately defined constructs, and (iii) recommend language improvements which reduce lack of expressivity, ambiguity, and vagueness (Almeida & Guizzardi, 2013). We refer to the ontology that analyzes a DSML as the reference ontology.
Over the last 15 years, a growing number of DSMLs have been analyzed using different types of reference ontologies. For instance, the integrated process modeling grammar within the ARIS framework has been evaluated using the Bunge Wand Weber (BWW) ontology by Green & Rosemann (2000), or the ArchiMate enterprise architecture language has been evaluated by the Unified Foundational Ontology (UFO) (Azevedo et al., 2015). Other ontological analyses of DSMLs were also performed on, for example, the RM-ODP language (Almeida, Guizzardi, & Santos, 2009) and the REA enterprise modeling language (Geerts & McCarthy, 2003).