Ontologies · Industrial data
Industrial data ontology
An industrial data ontology is a typed model of the data an industrial operation already holds: each asset, record, event, site and person becomes an object with properties and status values, anchored in the system of record that holds its data, and joined to the others by typed links. CodeNinja Atoms has published 28 industrial data ontologies, one per real operation across 11 sectors in the United States (14), Saudi Arabia (7), Pakistan (6), Germany (1): 397 objects, 1424 properties and 360 links, 328 of the 397 objects anchored in a named source system. Each is hyper-ontology/1 JSON under CC BY 4.0.
All 397 objects are one table, objects.jsonl, in the dataset; each design's model downloads as objects.json from the table below; the format is hyper-ontology/1.
What an industrial data ontology holds
Every object has a kind. Across the 28 published models these are the kinds, most common first, with real objects from different operations.
| Kind | Objects | What it is | Examples from the published models |
|---|---|---|---|
| asset | 112 | equipment and other physical things the operation owns and maintains | Formed structural part, Commuter car, Operating Facility / Site |
| record | 98 | what the operation writes down: work orders, tickets, findings, calibrations | Incident/Accident Record, Batch ticket, Failure mode |
| event | 50 | something that happens at a time: an alarm, a fault, a mission | Flight Mission, Condition Anomaly Alert, Active fault |
| site | 30 | a place the operation runs: a station, a field, a centre | Compressor Station, Distribution Center, Farm holding |
| actor | 24 | a role that acts on the operation: a desk, a crew, an officer | Mechanical desk controller, Field crew, Revenue field officer |
| document | 22 | a report, letter or dataset the operation keeps | QA/QC Accuracy Report, RCA report, OEM Authorization Letter |
| person | 20 | a named job a person holds: the one who reads, decides or approves | Shift Lead, Instrumentation Technician, Worker |
| measure | 20 | a reading, sample or score taken from the operation | Monitored point, Detector Reading, Criticality index |
| system | 14 | a software or control system the model names as an object in its own right | HVDC Transmission System, Edge Gateway, Data Source Connection |
| material | 7 | stock that moves through the operation: parts, wafers, containers | Spare part, Wafer, SKU |
The parts of an industrial data ontology
| Part | What it is | In the published models |
|---|---|---|
| Object | one thing the operation tracks, with a kind | 397 objects across 28 models |
| Properties | what the operation records about the object | 1424 properties |
| Anchor | the system of record that already holds the object's data; the ontology reads it and never replaces it | 328 of 397 objects name one |
| Status vocabulary | the fixed values the object's state can take | 127 of 397 objects carry one |
| Typed links | typed relations to other objects in the same model; what makes it an ontology rather than a table | 360 links, 336 of them named |
Every published industrial data ontology
Newest first. One model per real operation; open the design for the paper, or download the model as JSON.
The format
Every model is the same package, hyper-ontology/1: objects with kind, anchor, properties, status vocabulary and typed links. It is documented at hyper-ontology/1, loads with the Python loader, and stands up as a living ontology over the operator's own systems in Hyper Ontology.
Built from the published object models only (dataset objects.jsonl and designs.jsonl). Each object model, browsable object by object, at industrial ontology examples; the designs they come from at sovereign AI reference architectures, as an industrial AI reference architecture by sector and as physical AI system design. Designed on Praxis. Updated 2026-11-03.
Want one of these designs for your own operation? Talk to us.