Open Semantic Ecosystem for Enterprise & IT Architecture
Turning distributed architecture sources into reusable knowledge, enabling ontology-guided conversational analysis.
The Problem
ArchiMate models sit in Archi. Service catalogues live in Backstage. Decisions are buried in Confluence, infrastructure in Terraform, delivery context in Jira. Nobody can answer a cross-cutting question without days of manual research.
The answer isn't another mega-tool. Distribution is the natural state of architecture knowledge — each discipline has legitimate reasons for its own notation and system of record. Consolidation attempts fail because they fight that reality.
Linked.Archi makes the data model the integration point instead of the tool. Sources keep their native semantics and stay where they are; a shared semantic contract lets them be queried as one graph.
At its core is an ontology-driven framework for semantic understanding and decentralised interaction with Enterprise Architecture processes and artefacts — across domains, disciplines and tools.
By capturing the relationships between low-level technical information and high-level architecture, it lets questions be asked in natural language and answered from evidence held in the knowledge graph.
The Semantic Layer
Not interchangeable, and not a stack you climb in order — OWL and SHACL are independent concerns operating under different assumptions.
Classes, properties and axioms define what an Application is and how it
relates to a Capability. Open World — inference extends what is known.
Controlled vocabularies, navigation and semantic indexing — so everyone means the same thing by the same term, and concepts can be browsed and mapped.
Executable constraints, validation and derivation rules over architectural data. Closed World — governance that runs in CI, not in a review meeting.
Ontology-Guided Conversational Analysis
The ontology constrains interpretation, so terms resolve to real concepts and every claim traces back to triples in the graph — no hallucinated entities. Delivered through AI agent skills distributed via APM (Agent Package Manager).
What You Get
Ontologies, taxonomies, shapes, viewpoints, deliverable templates and presentation models.
Use the published metamodels as-is, extend them, or compose your own from the core.
Turn native model files into RDF — available as JAR and Docker for pipeline use.
Produce documentation, diagrams and stakeholder deliverables from the graph.
Agent Package Manager skills that let assistants query and reason over your graph.
Modelling guides, design decisions, cross-language mappings and validation practice.
Coverage
Each formalised as an OWL ontology with SHACL shapes and SKOS taxonomies, plus cross-language mappings between them.
Open source and standards-based. Start with the guide, explore the ontology index, or get in touch about applying it in your estate.