I think the OWL2 spec clarifies that imports should name the ontology or
version IRI, but in practice there are a lot of models that import by
URL. If an ontology document in a repo goes through a series of updates
and commits, some of which are tagged as having an updated version IRI,
the ontology developer might just give the users URLs for specific
commits to import the version they desire.
But does a DocumentGraphRepository care whether the addMapping string is
an ontology IRI or not? Isn't it just a URI that gets mapped to a URL
(the documentPath)? Couldn't I just do three different addMappings for
the same documentPath, and it doesn't matter whether the different users
import via URL, ontology IRI, or version IRI?
On 8/8/2024 2:07 AM, Sergei Zuev wrote:
In the new API, the ontology identifier is either the version IRI or the
ontology IRI. The document IRI is not an identifier.
Perhaps this is a flaw - document support was added at the last moment, and for
managing imports closure no document-manager is needed.
Typical work with documents can be like this:
```
DocumentGraphRepository repository =
GraphRepository.createGraphDocumentRepositoryMem();
repository.addMapping(
"http://www.co-ode.org/ontologies/pizza/pizza.owl";,
documentPath);
OntModel m = OntModelFactory.getModelOrNull(
"http://www.co-ode.org/ontologies/pizza/pizza.owl";,
OntSpecification.OWL2_DL_MEM_BUILTIN_RDFS_INF,
repository);
m.write(System.out, "ttl");
```
Here, `http://www.co-ode.org/ontologies/pizza/pizza.owl` is an ontology IRI.
`DocumentGraphRepository` manages mappings between documents and ontologies.
If you think that some functionality is missing, please open a github issue.
By the way, the external library owlcs/ontapi, which works on top of
apache-ontapi, has its own full-fledged ontology manager, which works according
to the specification, i.e. the ontology identifier can also be a document IRI.
On 2024/08/07 11:15:58 Steve Vestal wrote:
I have some questions about doing an OWL import closure. My
understanding is that an Import for an owl:Ontology can use any of a
URL, an ontology IRI, or a version IRI to access an OWL document.
Different documents that have the same ontology IRI may be accessed
using different URLs, either identical copies or with different version
IRIs.
The link given on the page
https://jena.apache.org/documentation/ontology/#graphrepository
(https://jena.apache.org/documentation/javadoc/jena/org.apache.jena.ontapi/org/apache/jena/ontapi/GraphRepository.html)
gets a "URL was not found" error. A search didn't turn it up. What
package contains this class?
That page says "By default, when an ontology model reads an ontology
document, it will/not/locate and load the document’s imports." A search
of this page did not find OntDocumentManager, which I have been using.
FileManager seems to be deprecated in favor of RDFDataMgr. Will
OntDocumentManager be deprecated or modified?
What is the recommended way to accumulate a set of URLs, ontology IRIs,
and version IRIs, and use that set to do an import closure?
