How can enterprise modelling tools be designed to support rapid, flexible visual language definition without requiring programming skills or costly redevelopment cycles?
Visual languages are central to enterprise modelling — from ArchiMate and BPMN to UML and custom domain-specific notations — but the tools that support them have a persistent problem: they are built around fixed meta models and hard-coded notations, making it difficult to adapt them for new purposes, different stakeholder groups, or evolving modelling needs. The result is a chronic mismatch between what tools offer and what practitioners actually need, compounded by the high cost and skill requirements of extending or replacing those tools.
This paper addresses the problem at its root by presenting an advanced meta-meta model — the layer that governs how concepts, relationships, properties, and visual representations are defined within a modelling environment. The model supports arbitrary meta model definition, multiple simultaneous visual languages for the same semantic model, rich property types, multi-level modelling, run-time extension without coding, and polymetric diagramming (where visual properties like size and colour reflect underlying data). It targets a property graph implementation, which offers a more natural fit for the richly interconnected structures of enterprise modelling than traditional relational or object databases. The design draws critically on two decades of experience with the EVA toolset — cataloguing what works well and what its architecture cannot support — alongside a systematic review of Eclipse EMF, MetaEdit+, XModeler, RDF/OWL, and property graph systems.
For researchers and tool builders working on the next generation of enterprise modelling environments, this paper provides both a rigorous theoretical foundation and a practically motivated design.
Originally published as a journal article by Graham McLeod in the EMISA Journal (Enterprise Modelling and Information Systems Architectures), with a companion presentation at the Models at Work stream, PoEM 2022.