Re: Provenance for section 3 in technologies.tex

Andy has already covered the key points well.

In particular, whilst Mark is right that several RDF processors use quads (at
least internally) they have completely different uses for the fourth element. It
is used for the identity of the stating (essentially reification), for
referencing a source document (provenance) and for holding an "is-true-in"
context. Even amongst the latter uses there are many different notions of
"context" as pointed out by Alberto. To take an extreme example, the sort of
fine gained "set of assumptions" contexts needed by reasoning engines like
planners are completely different in semantics and implementation from the "bag
of hidden statements" context used in things like CWM and I don't think either
of them is very much like what Sowa means by context.

Dave

"Seaborne, Andy" wrote:
> 
> Mark,
> 
> "Quads" is a term used for several different things.
> 
> > So although
> > they are non-standard at the moment, people are using them, so should we
> > really rule them out?
> 
> Jena could be said to use quads to store reifications.  A statement is
> stored with an "is reified flag" and it optimizes some uses of reification
> (particualrly statement, not stating, usage).  We have users who have large
> datasets who use this internal compression with database-stored statements.
> It is invisible at the API.
> 
> There are operations directed at helping working with reification - these
> are independent of the storage optimization.  We have had both
> reification-optimizing and non-optimizing store for the same API.
> 
> The RDF/XML generated is as per the working group.
> 
> They are not the same quads as used by RDFStore where the 4th component is a
> URI, and indicates the source of the statement.  This is visible at their
> API (and in their version of RDQL) and can not be encoded directly in
> RDF/XML.
> 
> The formulae in cwm/N3 are another approach again.  These are quoted graphs
> at the nodes of the containing graph (i.e. a new node type).
> 
> Using non-standard features precludes interoperability.
> 
> We need a concrete use case for provenance across systems (not just for
> managing data within SIMILE).
> 
>         Andy
> 
> -----Original Message-----
> From: Butler, Mark [mailto:Mark_Butler@hplb.hpl.hp.com]
> Sent: 26 June 2003 13:12
> To: www-rdf-dspace@w3.org
> Subject: RE: Provenance for section 3 in technologies.tex
> 
> Hi Dave
> 
> > Non-standard extensions would be best avoided if you want
> > SIMILE to be a full
> > participant in the semantic web.
> 
> But to take this back to my original suggestion does this apply to quads? My
> understanding from Andy is that they are used by RDFStore and a number of
> RSS processors, and from Jeremy that although Jena 2 does not have a quads
> API it does actually use a quad data structure "under the hood". So although
> they are non-standard at the moment, people are using them, so should we
> really rule them out?
> 
> I'd be interested in feedback here from Eric Miller and David Karger also?
> 
> thanks
> 
> Mark

Received on Thursday, 26 June 2003 12:06:43 UTC