Re: nomenclature in current document - ISSUE-65

You had stated the preference for your own document many times before. I 
don't think it is helpful at this stage to swap the meaning of terms 
such as Shape and Constraint. Doing such a renaming would likely push us 
back for several months, and make the whole discussion archive and 
history unusable.

Holger


On 8/14/2015 13:43, Peter F. Patel-Schneider wrote:
> Here is my current preferred nomenclature.  This has been excerpted from
> https://www.w3.org/2014/data-shapes/wiki/Shacl-sparql with minor changes
> (Core->Global, Extended->Scoped) and rearranged into a form that emphasizes
> the nomenclature.  Several details from the original account have been
> elided.
>
> This only covers nomenclature and some high-level aspects of syntax and
> validation.  There is not just a change in naming of things from
> http://w3c.github.io/data-shapes/shacl/ but also differences in how shapes
> and constraints are put together and in how validation works.
>
>
>
>
> SHACL constraints:
>
> Some SHACL constraints separate a constraint into a scope section and a
> shape section. These SHACL constraints, called SHACL Scoped Constraints,
> must have precisely one scope section and one or more shape sections.  Both
> the scope section and the shape sections are SHACL shapes.
>
> Other SHACL constraints have instead a SPARQL query and are called SHACL
> Global Constraints.
>
> All SHACL constraints have as well a severity level.
>
>
> SHACL shapes:
>
> There are multiple kinds of SHACL shapes, each with different syntax.  Some
> SHACL shapes have a component that is another SHACL shape.
>
>
> Validating a SHACL shape against a node in an RDF graph (in an RDF dataset):
>
> The result of validating a SHACL shape against a node in an RDF graph (in an
> RDF dataset) is zero or more shape violations.  If no violations are in the
> result then the validation succeeded, otherwise it failed.
>
> The different kinds of SHACL shapes each place different requirements on the
> "shape" of an RDF graph in the neighbourhood of the node.  The SHACL shapes
> that have a component that is another SHACL shape generally succeed or fail
> based on whether validating this other shape on one or more nodes in the
> graph succeeds or fails.
>
>
> Validating a SHACL Scoped Constraint:
>
> A SHACL scoped constraint is validated against an RDF graph (in an RDF
> dataset) by first determining for which non-literal nodes in the graph the
> constraint's scope succeeds.  Then each of the constraint's shapes are
> validated against each of these nodes.  The validation succeeds as a whole
> if each of these validations succeed and fails otherwise.
>
>
> Validating a SHACL Global Constraint:
>
> A SHACL Global constraint is validated against an RDF graph (in an RDF
> dataset) by evaluating its query against the graph (in the dataset).  If the
> query result is empty the validation succeeds and otherwise it fails.
>
>
> Validating a SHACL constraint graph against an RDF graph (in an RDF dataset):
>
> A SHACL engine takes as arguments an RDF graph containing SHACL constraints
> and an RDF graph (in an RDF dataset) containing information on which to
> validate the constraints.  A SHACL engine takes each node in the constraint
> graph that is an instance of sh:Constraint and validates the constraint
> encoded by this node against the information RDF graph.  The validation
> succeeds if no constraint with an error severity level fails and fails
> otherwise.
>
>

Received on Friday, 14 August 2015 06:32:49 UTC