Re: type and instance and subclass in SHACL documents

On 13/03/2016 8:10, Peter F. Patel-Schneider wrote:
> I believe that the working group would work better if most of its members
> understood the languages used in SHACL.  I had thought that such knowledge was
> common in the working group, but maybe not.

I find this a remarkable statement. FWIW I have 13 years of very intense 
RDF implementation experience, have a PhD in Computer Science and 
authored a paper that recently won a 10 years most influential paper 
award at ISWC. The company that I started the product line of now has 
almost 30 employees and is doing well. Yet I struggle to parse this 
discussion. The differences that you are referring to are simply *not 
practically relevant at all*.

Not accepting this reality will lead to yet another overly-theoretical 
standard that will fail in the marketplace. We, as a community, can do 
better, and deliver something that solves real-world problems. For the 
real world, RDF is just a data structure. RDF URIs are used in triples. 
There is no inherent special treatment of certain triples only because 
they use a certain namespace. If RDFS wants to apply a special treatment 
then good on them, and people can activate their RDFS reasoner as part 
of their tool chain. They can also apply an OWL reasoner - why limit 
this to RDFS? SHACL can handle that situation but does not depend on 
RDFS and only shares some of the RDFS URIs because otherwise we would 
end up with two separate semantic webs.

Holger

Received on Sunday, 13 March 2016 23:08:07 UTC