Skip to content

Ontochoros

Lint, test, diff, and CI for OWL ontologies — think dbt, for ontology engineering.

Ontochoros turns an OWL 2 ontology into an embedded, Cypher-queryable graph, then puts a standard engineering workflow on top — build, query, lint, test, diff, migrate:

ontopoiesis build pizza.owx -o pizza.lbug   # OWL/XML → queryable projection
ontopoiesis lint pizza.lbug                 # structural quality gates
ontopoiesis test pizza.lbug tests/          # your own Cypher checks, via pytest
ontopoiesis diff v1.lbug v2.lbug            # what actually changed, per axiom
ontopoiesis export pizza.lbug               # losslessly back to OWL/XML

Everything is pure Python: no JVM, no server, no specialized ontology API. If you can write a graph query, you can interrogate, test, and author an ontology. Run against widely used vocabularies — FoaF, the SNOMED CT ontology, the Human Phenotype Ontology — the stock lint rules surface structural defects in the released files that standard reports do not flag. Why this exists, and the evidence →

One graph, one query language

The build turns your ontology into a graph whose shape is OWL 2 — the structural specification, not the RDF encoding, so there are no reified axioms, rdf:first/rdf:rest chains, or blank nodes to reassemble — and any question about the ontology can be asked directly. Which live classes are defined using a term the file itself marks obsolete? is a single Cypher query, and it answers in seconds on graphs of a million-plus nodes. Lint rules, tests, diffs, and migrations all read that same graph: once you can query an ontology, you can check it, change it, and gate it in CI.

The packages

Three packages form one stack. Most people only need the CLI.

I want to… Reach for Start at
Lint, test, diff, and gate an ontology from the command line Ontopoiesis Quickstart
Query an ontology with Cypher inside my own Python Ontoplexis Quickstart
Move typed OWL 2 records between systems as JSON Ontophora Introduction

  • Ontopoiesis — the CLI you run

    Build, query, lint, test, diff, migrate, and render ontologies, locally and in CI.

  • Ontoplexis — the graph underneath

    Lossless OWL/XML ⇄ property-graph mapping as a Python library, for when you want raw Cypher in your own code.

  • Ontophora — the typed records

    Standalone Pydantic models for OWL 2 constructs, with a stable JSON format for moving them between systems.

Where to start