Re: ACTION-420 Review of SW-compatibility

Jos de Bruijn wrote:

> I just see three options:
> 1 define compatibility only with respect to standard RDF
> 2-define compatibility with respect to generalized RDF
> 3- define compatibility with respect to "semi-generalized" RDF, in which 
> you would not have literals and blank nodes in property positions
> 
> Option 3 does not make any sense to me, since I don't see any argument 
> for disallowing literals and blank nodes in predicate positions when 
> generalizing RDF.

My memory is that the RDF Core WG explicitly considered whether it could 
generalize the RDF data model and it is the case of blank nodes in 
predicates positions that caused problems for the model theory and 
caused it to be shunted into the "future work" bucket.

If that is relevant that the discussion on it will presumably be 
accessible in the archives or I could ask a colleague who was there but 
not this week.

> Option 1 has the potential advantage that it might be easier to 
> understand, because some people might not grasp the idea of generalized 
> RDF.
> Option 2 has the advantage that it can be used with an extended notion 
> of RDF graphs; it can does accommodate certain possible future extensions.
> 
> 
> As I stated earlier, it makes sense to me to consider generalized 
> graphs, because the syntactical restrictions imposed by RDF on triples 
> are rather silly.  However, I do not feel that strongly.

It makes sense to me too. As I've mentioned before Jena implements 
generalized graphs underneath and the rule engine works at this level 
then restricts the outputs to match RDF constraints.

However, I don't feel that strongly about it either.

Dave
-- 
Hewlett-Packard Limited
Registered Office: Cain Road, Bracknell, Berks RG12 1HN
Registered No: 690597 England

Received on Monday, 10 March 2008 19:21:19 UTC