GSQ vocabs must be valid according to the VocPub Profile which is a specification that constrains SKOS to ensure that certain vocab element properties are present, such as definitions for Concept instances, basic ConceptScheme metadata etc.
In addition to all the rules imposed by VocPub and, through inheritance SKOS, GSQ vocabs must also be valid according to rules within the GSQ Profile of VocPub which simply adds a few more GSQ-specific vocab tests.
Vocabulary validation is performed technically using graph pattern matching data objects called shapes. These are like a template of content that must be present or configured in a particular way in order for the vocabulary to be valid according to some specification.
VocPub provides a validator written in the SHACL constraint language which can be downloaded here:
A working copy of it is stored in this directory as vocpub.ttl too.
The extended GSQ validator is stored in this directory as gsq.ttl.
A further validator, gsq-compounded.ttl, is a combined validator including the GSQ Profile of VocPub's validator, the VocPub Profile validator and a SKOS validator all in one. It is all you really need to validate GSQ vocabs with.
To use those GSQ and other validators, you need a SHACL engine that can apply the validator file to a target data file. Here are options for doing that.
You can use the RDF Tool to validate RDF data online. It comes pre-loaded with many validators and you can also add your own in too. Just select your validator, or upload / copy 'n paste it, and your data file in the same way and hit validate.
It will be updated soon with this new VocPub validator version and with the GSQ validator.
Use the Python pySHACL tool which is the basis for the RDF Tools validation service above, on the command line, something like this:
~$ pyshacl -s vocpub.ttl ../vocabularies/borehole-design.ttl
-s indicates the shapes validator file and ../vocabularies/borehole-design.ttl is the vocab under test.
Any validation errors will be printed to screen.
All vocabs can be validated at once using the bulk validator Python script validate_all.py which uses pySHACL. Run it like this:
python validate_all.py report