Re: prospective charter for a shapes validation WG

On 07/10/2014 06:07 AM, Sandro Hawke wrote:
> On 07/10/2014 08:32 AM, Peter F. Patel-Schneider wrote:
>> I find part of Sandro's motivation rather dangerous.  I do not think that
>> providers of RDF should be in the business of tailoring the shape of their
>> output for eventual consumption.  It is the business of consumers to decide
>> whether the RDF that they encounter is suitable for their purposes.
>>
>
> Very interesting.  That sounds like a business ignoring what customers want or
> a speaker never thinking about what the audience might want to hear.   Can you
> explain this a bit more?

Sure.

If as a speaker I tailor my words to exactly match the shapes that every 
audience member wants, I can only talk to very small audiences (less than or 
equal to 1, probably).  This is, of course, a speaker model, but often not a 
very good one.  If as a company I tailor my products to exactly match the 
shapes that every customer wants, I end up creating very many products 
(probably more variations of each product than I have customers of that 
product).  This is, of course, a business model that can be effective in some 
cases.

As a publisher of RDF should I follow this route?  I say no!  More, I say that 
this is against the principles of the Web.


This is not to say that there can't be situations where RDF is used in 
one-to-one settings, and it may be expedient for the producer to tailor its 
output to match the desires of the consumer.  However, these situations are 
likely to involve constraints on more than the shape of the RDF, for example 
requirements concerning precisely what is and what is not produced at a 
content level.

It seems to me that remote SPARQL execution is a better model for this sort of 
one-to-one interaction, and that this model is not against the principles of 
the Web as there is a publishable RDF graph that the queries are performed 
against.


>> That said, there are reasons to have a mechanism for checking and possibly
>> modifying RDF graphs.  This mechanism could be helpful in determining
>> whether a particular RDF graph carries sufficient information to be useful
>> for a particular consumer.
>>
>
> +1
>
>         -- Sandro

peter

Received on Thursday, 10 July 2014 14:43:53 UTC