best practice for rdfs:subClassOf closure

[danbri]

I'm working with @Gnomus042 on SHACL/ShEx topics and we had need to update a Graph to explicitly add all the implied rdfs:subClassOf arcs. We found two approaches, one using transitiveClosure(), and the other using SPARQL .query():

Working examples - https://colab.research.google.com/drive/1DWrR-VKus4elIJ2lpHy7BXw7Jhk5qmDH?usp=sharing

Can you advise on which is preferred, or whether there is some other built-in or nearby extension to perform such tasks (and related inference e.g. subPropertyOf etc). I searched around but couldn't find clear guidance. Thanks for any pointers!

[ashleysommer]

@danbri
The owl-rl project can operate in RDFS mode, which will expand a graph to inference all RDFS axioms including subClassOf and subPropertyOf. To answer your first question, I believe all graph expanders built on rdflib use the transitiveClosure() mechanism.

I see in your examples you're using PySHACL. Note, pySHACL has owl-rl built in, you can expand your graph before validation automatically using inference="rdfs" when calling validate().