use case for named g-snaps ?

On 08/23/2012 12:52 PM, Dan Brickley wrote:
> On 23 August 2012 17:00, Sandro Hawke <sandro@w3.org> wrote:
>
>> You lost me here, sorry.   What's the use case for an immutable named graph?

[ I'm going to go back to calling them "named g-snaps" for this 
discussion, since this thread is no longer talking about the retronym 
naming proposal. ]

> How about official snapshot releases of gigantic datasets (wikidata or
> freebase dumps, BTC, etc.). We could speak of 'full dbpedia dump as of
> 2012-05-01', regardless of where the triples were gotten (ntriples,
> caches, dvd/cdrom, bittorrent).

Hmmm.

The more common approach, I think would be to have several URIs that 
give you the same triples, like:

(u1)   http://dbpedia.org/dump/2012-05-01
(u2)   http://example.org/dbpedia-dump/2012-05-01
(u3) 
http://example.com/image/a6b7d2a971bf47ae22a413a29b681becaef89801b9782902bd275da116bde00a


And each of them would have some of their own metadata properties, like:

<mailto:webmaster@dbpedia.org> eg:responsibleFor 
<http://dbpedia.org/snapshot/2012-05-01>.

and

<mailto:webmaster@example.org> eg:responsibleFor 
<http://example.org/dbpedia-snapshot/2012-05-01>

And there would be stuff about when and how the mirror was made, that 
someone promised never to change it, etc.

In this case, u1, u2, and u3 denote gboxes.     (They can't denote 
g-snaps, because they contain the same triples but have different 
properties.  g-snaps can't do that.)

But you were suggesting something different.  You were suggesting:

(u4)   http://dbpedia.org/gsnap/2012-05-01

which *would* denote the g-snap.   So its properties would have to be 
properties of the gsnap, not of a gbox or service endpoint or 
whatever.   Triples using u4 would be triples about the gsnap, unrelated 
to any particular gbox which happens to be holding it.

So we could say things like how many triples it has, how much people 
like it, whether it's "true", etc.

I see some merit to that approach -- it's very clean modeling -- but I'm 
not sure it's worth it.   I think it would lead to a lot of confusion in 
the httpRange-14 space.  I think it's probably simpler to just say those 
things using u1, u2, or u3, using properties which talk about the 
triples in that gbox.  That way everything remains observable, more 
concrete.

     -- Sandro

> Dan
>
>> And it sounds like you're suggesting "mutable named graph" as the official
>> term for g-box.  Is that right?

Received on Friday, 24 August 2012 02:30:34 UTC