Skip to content

Related Work: Figay ArchiMate OWL Formalization

Nicolas Figay's ArchiMate OWL project is an experimental exploration of OWL reasoning over ArchiMate models — specifically, whether OWL DL axioms can infer and validate which relationships are valid between which element types.

Repository: github.com/nfigay/archimate.owl Author articles: Medium — @nfigay

Overview

Dimension Linked.Archi Figay
ArchiMate version 3.2 and 4.0 3.0
Primary goal Metamodel platform for architecture knowledge graphs Experimentation on OWL reasoning over ArchiMate models
Constraint enforcement SHACL shapes OWL axioms + DL reasoning
Relationships Three-declaration qualified pattern + rdf:reifies OWL ObjectProperties with OWL restrictions
Validation SHACL (relationship, element, principle, viewpoint shapes) OWL reasoning (consistency checking)
Status Active development Experimental, last updated ~2019

Context

Nicolas Figay is an enterprise interoperability expert. The repository is described as "experimentation on OWL derivations from ArchiMate, dedicated to usage of reasoning on imported ArchiMate models." It contains multiple OWL files exploring different approaches:

  • ArchiMate 3.0.owl — Core ArchiMate 3.0 formalization
  • ArchiMate 3 for inferring relations from Open Format exchange files.owl — Extended version with relationship inference rules
  • ArchiMate with Grakn experimentation.owl — Exploration with the Grakn/TypeDB knowledge graph platform
  • XSLT scripts for transforming ArchiMate exchange files

The project is exploratory — multiple approaches coexist without converging on a single canonical ontology.

The reasoning question

The central question Figay's project explores: can OWL DL axioms encode the ArchiMate relationship validity matrix so that a reasoner can determine which relationships are valid between which elements?

The answer is: partially. OWL can express some constraints (e.g., "assignment can only originate from active structure elements") via domain/range restrictions and class expressions. But the ArchiMate relationship matrix is a closed-world constraint — "these specific pairs are allowed, everything else is forbidden." OWL operates under the open-world assumption — the absence of a statement doesn't mean it's false.

This is exactly the tension that led us to adopt SHACL for validation (DD-2). SHACL provides closed-world validation: "if this relationship exists and its source/target types are not in the allowed set, report a violation." OWL cannot express this without making the ontology inconsistent (which is a different thing from invalid).

Figay's project is a useful demonstration of where OWL reasoning hits its limits for metamodel constraint enforcement.

What we learned from this project

  • OWL reasoning is the wrong tool for relationship validity. The ArchiMate relationship matrix is a closed-world constraint. OWL's open-world assumption means you can't say "only these pairs are allowed" — you can only say "these pairs exist." SHACL is the right tool for this.

  • The interoperability framing is valuable. Figay's articles on semantic cartography and linked enterprises articulate the problem well: different tools export different representations, and a canonical ontology is needed for interchange. This is the same problem Linked.Archi addresses, from a different angle.

  • Grakn/TypeDB exploration shows the problem isn't RDF-specific. The ArchiMate-to-knowledge-graph problem exists regardless of the underlying graph technology. The modeling decisions (how to represent relationships, how to enforce constraints) are more important than the platform choice.

  • Experimental projects have value even when incomplete. Figay's work demonstrates what doesn't work (OWL for closed-world constraints) as clearly as what does. That's useful knowledge for anyone building in this space.

References