Re: How to publish SPARQL endpoint limits/metadata?

Hmm.
In my mind, a dataset is rather abstract - a collection of data that is being made available.
They may use a combination of any or all of SPARQL endpoints, downloads of "dump"s, and resolvable URIs (Linked Data).
They may also make it available in other forms, but we are possibly primarily concerned with RDF here, although it would be a shame if we could not embrace the more abstract concept.

As a consumer (always!), I would like to come to where I think the dataset is being "published", or look in some aggregator index, and easily find out all the stuff I need to know about the dataset, and how I might use it.

That's my starting point.

So in our system like many others around I think, for example, when we get a new URI, we hope there is a SPARQL endpoint, as that is our preferred format.
We need to use internal information to do this, so we can only do it for known places.
If not, then we try to simply resolve it.
Failing that, we could look in a cache of dumps we have found, but don't actually at the moment.

It would be good, for example, if resolving the URI always told us where there is metadata about a SPARQL endpoint that is recommended as having RDF about this URI.
In fact, we do this for co-reference information for our URIs (we use a bespoke predicate, but should probably have been using seeAlso), but should probably do it for SPARQL endpoint as well.
The metadata should be at the end of a resolvable URI, and the SPARQL endpoint should hold its own metadata in it, etc. etc..

So having a separation between SPARQL Service Description and voiD would just be plain wrong.
They must embrace each other, so that consumers can easily work out how to use what they think of as a "dataset".

I would also add that if I take a REST-like view of the world, which I do for accessing a SPARQL endpoint (I am simply retrieving a document), the distinction between dataset and service becomes very blurred.
Even calling it a "SPARQL Service Description" seems rather old-fashioned to me.

Best
Hugh

On 9 Oct 2013, at 11:04, Barry Norton <barrynorton@gmail.com>
 wrote:

> On Wed, Oct 9, 2013 at 10:55 AM, Frans Knibbe | Geodan <frans.knibbe@geodan.nl> wrote:
> 
> Shouldn't that be the SPARQL Service Description instead of VoID? In my mind, SPARQL endpoints and datasets are separate entities.
> 
> 
> +1

Received on Wednesday, 9 October 2013 11:17:36 UTC