Ontology import chain

The basic principle that guides the structure of the import chain is that ALL references to owl:Classes, owl:ObjectProperties, etc. should not require a local redefinition of that class. Said another way, all local additions to externally defined classes should appear in the Annotations section of the serialization, not in the Classes section. It is not always possible (nor sensible) to follow this principle, but most of the time it is.

If there is an external class that we want to reference it should either be imported directly or it should live in a slim that is imported as a proxy for the external ontology and copies exactly the contents of that ontology. Any local modifications should NOT be made to that slim itself but should be made in the bridge file that imports it (in accordance with the basic principle).

In the case where we want to make bridging axioms between multiple external ontologies, those axioms should live in another bridge file that imports the other bridge files. This practice increases the depth of the import chain but ultimately simplifies the development workflow by providing explicit locations where bridging axioms should live that can be easily checked by seeing whether the basic principle holds.

Import of ttl/generated/*-dead.ttl files should occur at the first point in the import chain that we have control over. See for example doid-bridge.ttl which imports both symp-dead.ttl and trans-dead.ttl. The original ontologies are imported by doid.owl so we import them in doid-bridge.ttl.