Ontologies
An ontology defines the business concepts that live across your semantic models. Instead of thinking in tables and columns, you define entities - things your business tracks, like Customer, Product, or Invoice - and map them to the physical tables where they live. This gives AI tools and downstream consumers a shared vocabulary for understanding your data, even when the same concept appears in multiple models with different names.
Why Ontologies Matter
Ontologies are the foundation for Semantic Modeler's MCP (Model Context Protocol) server support. When an AI agent connects to your models through an MCP server, the ontology is what lets it understand your data in business terms - it knows that "Customer," "Client," and "Account" all refer to the same concept, which tables hold that data across your models, and how concepts relate to each other. Without an ontology, an AI agent only sees raw table and column names; with one, it can reason about your data the way your business does.
Ontologies also feed into the AI Readiness score in Model Health, and their structured export formats (JSON-LD, Turtle) make them interoperable with other semantic web and linked data tools.
Key Concepts
- Entities - business "nouns" with identity. Customer, Product, and Invoice are entities. If its records have a key someone could read over the phone ("Invoice 10437"), it's an entity.
- Hierarchies - entities can be linked into parent-child hierarchies using the Parent (level of) field. For example, three separate entities - Customer Parent, Customer Master, Customer - can be arranged into a hierarchy by setting each child's parent.
- Synonyms - alternative names for an entity. If your CRM calls a "Customer" an "Account," add "Account" as a synonym rather than creating a new entity.
- Bindings - where an entity physically lives: which model, which table, and which identity key column(s). Two bindings of one entity are the cross-model join - their key columns pair by rank.
- Relationships - connections between two different concepts, expressed as subject - verb - object (e.g., "Product - sent to → Customer Master").
- Member Models - the models included in this ontology. Only checked models are scanned for entities and relationships.
Review Workflow
Each entity has a Needs review toggle. Use it to track which concepts have been validated by the business and which still need sign-off. A business analyst can work through the entity list, toggling the flag off as each definition, binding, and set of synonyms is confirmed - so the team always knows what's been approved and what hasn't.
AI Scan
The AI scan analyzes your member models and proposes entities and relationships. Scans only ever suggest - nothing is added until you accept it.
Export and Import Formats
| Format | Description |
|---|---|
| XLSX | Excel spreadsheet. Use this to copy ontologies between instances of Semantic Modeler, or hand it to a business analyst to review and edit entities, synonyms, and bindings in Excel. Import the modified file back to apply changes. |
| ODS | Open Document Spreadsheet. Same purpose as XLSX for teams using LibreOffice, OpenOffice, or other ODS-compatible tools. |
| JSON-LD | JSON for Linking Data - a W3C standard that embeds semantic meaning into JSON. Best for programmatic integration with APIs, data catalogs, and other systems that consume linked data. Human-readable and widely supported in web tooling. |
| Turtle (.ttl) | Terse RDF Triple Language - a compact, human-readable RDF serialization. Preferred in semantic web workflows, SPARQL endpoints, and knowledge graph tools. More concise than JSON-LD for complex ontologies with many relationships. |
| Documentation (HTML) | A human-readable reference document rendered as HTML. Useful for sharing with stakeholders who don't use Semantic Modeler. |
| Documentation (Markdown) | Same reference content as HTML, but in Markdown format for embedding in wikis, READMEs, or documentation sites. |
XLSX and ODS files can be imported back into Semantic Modeler via the Import... option in the ontology context menu, making them the primary format for round-trip editing and transferring ontologies between environments.
User Guide
For a step-by-step walkthrough of creating an ontology, adding entities and bindings, and defining relationships, see Creating an Ontology.