Re: isolating shapes in named graphs

On 11/26/2014 8:14, Eric Prud'hommeaux wrote:
> I disagree that the majority of shapes would be global invariants. 

This depends on the use cases:

1) If the user wants to publish an "ontology" describing classes and 
their intended semantics, then I guess the constraints are meant to be 
global invariants.

2) If the use case is to identify instances that match certain patterns 
among sub-sets of all instances then the notion of stand-alone shapes 
can be helpful syntactic sugar, as an alternative to matching those 
instances in SPARQL directly.

SPIN constraint checking so far has focused on scenario 1) while leaving 
scenario 2) to "normal" SPARQL operations. For example we have many SPIN 
templates based on SELECT queries that find instances with certain 
conditions.

Holger

Received on Tuesday, 25 November 2014 23:38:55 UTC