Representing instances: the case for reengineering conceptual modelling grammars

ABSTRACT

While many conceptual modelling grammars have been developed since the 1970s, they share the general assumption of representation by abstraction; that is, representing generalised knowledge about the similarities among phenomena in a domain (classes) rather than about domain objects (instances). This assumption largely ignores the fundamental role that instances play in the constitution of reality and in human psychology. In this paper, we argue there is a need for a grammar that explicitly recognises the primary role of instances. We examine the limitations of traditional class-based approaches to conceptual modelling, especially for modern information environments. We then explore theoretical and practical motivations for instance-based modelling, and show how such an approach can address the limitations of traditional modelling approaches. We conclude by calling for the engineering of instance-based grammars as an important direction for conceptual modelling research to address the limitations of traditional approaches, and articulate five challenges to overcome in such efforts.

ACCEPTING EDITOR:

• Par Agerfalk

ASSOCIATE EDITOR:

• Jan Marco Leimeister

KEYWORDS:

• Conceptual modelling
• information systems analysis and design
• requirements elicitation
• database design
• information quality
• ontology
• cognition

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. In this paper, we use the term “model” to refer to the output of conceptual modelling. Models are also known as diagrams, representations, schemas or scripts (Olivé, 2007; Peckham & Maryanski, 1988; Wand & Weber, 2002).

2. Conceptual modelling research typically uses terms such as (1) classes or entity types, (2) instances, objects, members and entities, and (3) attributes to denote what psychology may refer to as (1) concepts, categories, and classes (2) objects, individuals and (3) features, attributes or properties, respectively. These terms in philosophy are, correspondingly, (1) kinds, universals, sets or classes, (2) things, particulars, objects or individuals, and (3) properties, qualia, or features. For consistency, in this work, we use terms classes and instances when discussing issues related to IS and conceptual modelling, but also use other terms (e.g., objects, individuals) when it is more appropriate to the context or references discussed.

3. See, http://pubs.opengroup.org/architecture/archimate3-doc/chap09.html.

4. One workaround to deal with this challenge is to create a class for each new instance, but this would result in the vast increase in complexity of models and proliferation of classes, which would negate the benefits of class-based modelling. Furthermore, this would result in the confusion between what is a class and what is an instance.

5. A set of necessary and sufficient conditions that definitively determined whether a particular thing in a domain is a member of the class.