RE: Semantic content negotiation (was Re: expectations of vocabulary)

--Danny,

> That sounds like a "slippery slope" argument. I'm not 
> suggesting an XPath cookie, just something specifically to 
> enable useful RDF graph partitioning.

I agree with your concern.  Big RDF document is a big problem and must find
a way to resolve it.  But I think, first, the ontology designer should think
of how to partition their content.  Not only for the sake of network
bandwidth but also the ontology sharing, reuse and evolution.  Second, the
semantic cookies can be conviniently solved at the RDF level.  For instance,
if I have ontology with three partitions or closures.  My ontology could
easily be deployed something like this:

<> closure:type1 uri1 ;
   closure:type2 uri2 ;
   closure:type3 uri3 ;
etc...

With closure:typeN being a rdfs:subPropertyof owl:import.  Hence, when a
client retrieve this set of RDF, it can decide according to the hypothetical
closure ontology.  Wouldn't this a much convinient and clean way to solve
your problem?  

Hmm... maybe, we should do this and develop this closure ontology? I will
definetly start chewing about it.  If anyone interested, let me know.

Cheers,

Xiaoshu