On 12-02-07 06:50 PM, Armando Stellato wrote:
Dear all,
a simple question about publishing datasets according to the VOID
specification:
I've a SKOS concept scheme (in the specific: AGROVOC:
http://aims.fao.org/aos/agrovoc/) to publish according to the VOID
specification. One very simple thing which came to my mind was: why not
using the baseURI (again: http://aims.fao.org/aos/agrovoc/) of the
scheme as a resource and publish it as a DataSet?, much like the common
practice in ontologies is to use the baseuri as the resource identifying
the ontology itself. This way, I would not use any file, and I would
just make the data accessible through the SPARQL endpoint. The baseuri
would also return - through HTTP access, in case of a request for any of
the RDF mime-types - exactly the description of the Dataset.
This is compliant with many access modalities suggested in the void
guide (e.g. discovery of dataset through SPARQL queries would still find
the dataset declaration).
However, I did not find any example like this in the DERI guide nor in
the W3C draft, so was wondering if there is any reason for rejecting
this possibility.
Best,
Armando Stellato
Hi Armando,
The DataGovIE's VoID [1] describes the graph names for the vocabularies
that are used in the datasets with the SPARQL Service Description.
If your SPARQL service is able to dereference remote graph IRIs (using
FROM/FROM NAMED), you could get them on the fly. However, that's not an
ideal solution. A local copy of the vocabulary is more preferable in my
opinion.
[1] http://data-gov.ie/void.ttl
-Sarven
