W3C home > Mailing lists > Public > semantic-web@w3.org > July 2006

Re: baseURI - SPARQL endpoint - Named Graph

From: Richard Newman <r.newman@reading.ac.uk>
Date: Wed, 5 Jul 2006 10:57:38 -0700
Message-Id: <558B2F26-61A2-45A0-9A39-ACFACE14B8D1@reading.ac.uk>
Cc: "'SW-forum'" <semantic-web@w3.org>, "'Paap, Onno'" <onno.paap@gmail.com>, "Benjamins, Robin" <rxbenjam@bechtel.com>
To: "Hans Teijgeler" <hans.teijgeler@quicknet.nl>
> The data in a quad store has been mapped from some application to  
> the ISO 15926 format, and is owned by the owner of the quad store.  
> All other data must be fetched from the other stores.
>
> Each quad store has its own SPARQL endpoint, and its own base URI  
> (but we don't know whether that base URI is necessary for that  
> endpoint or not)

The base URI in SPARQL is a syntactic convenience;

BASE <http://example.com/>

... <horse> ...

is the same as saying

... <http://example.com/horse> ...

.

I advise you to forget about base URIs.

> RAP supports Named Graphs, and now the plot thickens.... How does  
> that relate to the endpoint and the base URI of that same store?  
> For your information: we intend to (mis?-)use the Named Graphs for  
> a compartimentation of the quad store, e.g. import archive, export  
> archive, population archive, handover archive, and of course the  
> actual 'payload' information content.

They don't relate. Your SPARQL queries can reference any named graph  
you like, and the default graph is unnamed. The endpoint, similarly,  
can be any URI you like*, and the base URI can simply be a useful  
part of a URI to shorten your queries.

(If you wish, you can somehow alias a name, which could be the same  
as the endpoint, to the default graph, but I don't see the point.)

> One last question: is there a mandatory format for SPARQL  
> endpoints? Can these two endpoints co-exist:
> http://www.pqr-ltd.com/p4502/ and
> http://www.pqr-ltd.com/p4502/me03  ?
> Is that last slash '/' required?

As far as I know there is no mandatory format. If the two resources  
reachable by the above URIs when dereferenced can behave differently  
(which they can), then I see no problem with it. I wouldn't imagine a  
trailing slash being necessary; both are valid absolute URIs,  
reachable over HTTP.

HTH,
-R

* With the obvious restriction that you need to put a SPARQL service  
behind it!
Received on Wednesday, 5 July 2006 17:57:58 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 21:45:11 GMT