How can visual modelling languages in enterprise architecture be designed to genuinely communicate meaning — rather than confusing or alienating the stakeholders they're meant to serve?
Graphical models are everywhere in enterprise architecture — yet a persistent gap exists between the effort invested in building them and the value they deliver. Models are too technical for business audiences, too homogenous to highlight what matters, or presented in formats that stakeholders simply cannot parse. When practitioners try to bridge this gap by converting rigorous models into PowerPoint slides or Word documents, they sever the connection to the underlying repository — destroying integrity, reusability, and currency in the process.
This paper presents the research programme Graham McLeod is pursuing at the University of Duisburg-Essen, supervised by Prof. Ulrich Frank, to address these problems at a foundational level. The research draws on human visual cognition, semiotics, information encoding theory, the Physics of Notations, and the emerging field of polymetric diagramming — a technique that modifies visual symbol properties such as size, colour, and shape to reflect underlying data, enabling pre-attentive processing and rapid identification of important patterns in large, complex models. The proposed contributions include extended theory for visual notation design, a meta-meta model supporting multiple visual languages over the same semantic model, and a layered tool architecture enabling runtime adaptation of models to purpose, audience, and medium.
For enterprise architects, this research points toward a future where modelling tools can produce representations genuinely suited to a CFO, a process owner, or a technical architect — from the same underlying repository, without manual translation.
Pages: 16
Originally published as a doctoral consortium paper by Graham McLeod in the PoEM 2018 Doctoral Consortium Proceedings (CEUR-WS Vol. 2234), Vienna, Austria, 2018.