Re: SHACL for datasets

Thanks - the catalog is one place to put the shapes info as is the data 
itself.

For shapes related to ontologies/vocabularies in an RDF Dataset, with 
lots of named graphs, targeting them could be done by using terms from 
the name space and have the engine "pull" in the shapes base on use of 
the terms.  That's quite rigid - some graphs may be of higher quality, 
or be work-in-progress, so always apply shapes isn't what is needed.

     Andy

On 22/07/2019 11:46, Natasa.Varytimou@refinitiv.com wrote:
>
> In Refinitiv we heavily utilize both dcat and SHACL
>
> DCAT is part of our models in our Metadata Registry where Dataset has 
> complete provenance, licencing and rights information and we have its 
> Distributions and Distribution Services.
>
> All our models are ontologies with classes which are also node shapes 
> and are completely described with SHACL.
>
> The models are stored in Named Graphs in Neptune and will also be 
> stored in xdb ( the equivelant) in TopBraid EDG
>
Translation: xdb is a way of using Apache Jena TDB.

> The actual RDF data are again Named Graphs in  Neptune and xdb in 
> TopBraid EDG that*import* the SHACL files that describe and validate them
>
> So it is not the SHACL graphs that target the RDF datasets but the RDF 
> datasets that import the SHACL graphs
>
> Btw, we found more than necessary to have closed shapes where 
> inheritance of attributes are also in play and we are using the dash 
> vocabulary instead of the SHACL closed shapes. It will be nice to move 
> that to the SHACL standard
>
> *From: *Dan Brickley <danbri@google.com>
> *Date: *Sunday, 21 July 2019 at 23:35
> *To: *Andy Seaborne <andy@apache.org>
> *Cc: *"public-shacl@w3.org" <public-shacl@w3.org>, Richard Wallis 
> <richard.wallis@dataliberate.com>
> *Subject: *Re: SHACL for datasets
> *Resent from: *<public-shacl@w3.org>
> *Resent date: *Sunday, 21 July 2019 at 23:35
>
> yes, e.g. see https://github.com/SEMICeu/dcat-ap_shacl 
> <https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_SEMICeu_dcat-2Dap-5Fshacl&d=DwMFaQ&c=4ZIZThykDLcoWk-GVjSLmy8-1Cr1I4FWIvbLFebwKgY&r=rMkwH8pHm4nx5b60gthskHySmXToxdxBmse0URUsEGg&m=7fYe7Xthic9viT2f3wva7AuNplJw1vq2zCMeitxBKNU&s=lf-AwpAUyDvYUlAY_t7bxPoJqCx7ow3fJUhRSFqALcc&e=> also 
> https://github.com/SEMICeu/dcat-ap_shacl/issues/32 
> <https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_SEMICeu_dcat-2Dap-5Fshacl_issues_32&d=DwMFaQ&c=4ZIZThykDLcoWk-GVjSLmy8-1Cr1I4FWIvbLFebwKgY&r=rMkwH8pHm4nx5b60gthskHySmXToxdxBmse0URUsEGg&m=7fYe7Xthic9viT2f3wva7AuNplJw1vq2zCMeitxBKNU&s=TMtuIO8xEeTyzUOvkaxpVfSVobacVGUgAKjcFt0xD3U&e=> 
> which explores ShEx and maybe schema.org 
> <https://urldefense.proofpoint.com/v2/url?u=http-3A__schema.org&d=DwMFaQ&c=4ZIZThykDLcoWk-GVjSLmy8-1Cr1I4FWIvbLFebwKgY&r=rMkwH8pHm4nx5b60gthskHySmXToxdxBmse0URUsEGg&m=7fYe7Xthic9viT2f3wva7AuNplJw1vq2zCMeitxBKNU&s=sYsn02qk3z_zJhsRqhPAgFCUgMs-AQTzugCIYNxB-JE&e=> 
> and dcat-ap variants, I forget. Am interested to go deeper into this 
> and there is potential Google interest too in this area
>
> Dan
>
> On Sun, 21 Jul 2019, 20:10 Andy Seaborne, <andy@apache.org 
> <mailto:andy@apache.org>> wrote:
>
>     Has there been any work on SHACL for RDF datasets?
>
>     Here are a some ideas:
>
>     https://afs.github.io/shacl-datasets.html
>     <https://urldefense.proofpoint.com/v2/url?u=https-3A__afs.github.io_shacl-2Ddatasets.html&d=DwMFaQ&c=4ZIZThykDLcoWk-GVjSLmy8-1Cr1I4FWIvbLFebwKgY&r=rMkwH8pHm4nx5b60gthskHySmXToxdxBmse0URUsEGg&m=7fYe7Xthic9viT2f3wva7AuNplJw1vq2zCMeitxBKNU&s=iXOuPOmjN5kGQdsM88pKm5K41MRaZy-g5xf7GQc0XJY&e=>
>
>     It has two parts -
>
>     1/ A ":targetGraph" to identify the data graph for the the shape,
>     including being able to say "all graphs", and "union of all named
>     graphs".
>
>     2/ A mechanism for having groups of shapes that can be
>     graph-targetted
>     together, rather than labelling a lot of shapes with the same
>     ":targetGraph".
>
>          Andy
>