Re: on determinism in SPARQL constraints

If SHACL was defined by a single shape mapping to a single SPARQL query then
the answer would be easy - whatever the query would do.  However, the current
processing model in SHACL is very unclear so there is no guidance as to the
order of processing of SPARQL queries or even how often they are processed.

I think that it is up to the working group to determine what sort of
processing model they want.  All I did was point out that without some sort of
mention of what processing models are allowed it is not possible to determine
the possible outcome of some shapes.

peter


On 10/02/2016 06:48 AM, Dimitris Kontokostas wrote:
> Hi again Peter,
> 
> This is an informal reply and not necessarily representing the views of the WG.
> 
> I looked a bit further at the SPARQL spec and I found these official
> functions: rand(), now(), UUID(), STRUUID().
> Of course, there may be other similar user defined functions.
> 
> Since you raised this comment, do you have any specific suggestions on how
> SHACL should work is such cases or a warning would be sufficient (for you)?
> 
> Dimitris
> 
> On Sun, Oct 2, 2016 at 11:39 AM, Dimitris Kontokostas
> <kontokostas@informatik.uni-leipzig.de
> <mailto:kontokostas@informatik.uni-leipzig.de>> wrote:
> 
>     Not necessarily, I could adjust my example similar to yours and run it
>     near the midnight of Oct 9th.
> 
>     In general all functions that return values independent of the data graph,
>     like rand or now may, in some cases, result in non deterministic
>     validation scenarios.
> 
>     Dimitris
> 
>     Typed by thumb. Please forgive brevity, errors.
> 
> 
>     On Oct 1, 2016 23:48, "Peter F. Patel-Schneider" <pfpschneider@gmail.com
>     <mailto:pfpschneider@gmail.com>> wrote:
> 
>         That's probably a separate issue as the different results would
>         (probably) be
>         produced by two different validations of a data graph.  In the case I
>         mention
>         below, the question is how often RAND is called during a single
>         validation of
>         a data graph.
> 
>         peter
> 
> 
>         On 10/01/2016 01:36 PM, Dimitris Kontokostas wrote:
>         > Thanks Peter,
>         >
>         > This is a good point but a possible constraint would be "an issue
>         should not
>         > be issued with a future date"
>         >
>         > given the following data graph:
>         >  ex:issue1 ex:issued "10-10-2016"^^xsd:date
>         >
>         > if we run the validation now we would get different results compared to
>         > running a validation with the exact same input after 10 days.
>         >
>         > How could we define a validation scenario to be deterministic but at
>         the same
>         > time allowing such cases?
>         > Requiring identical shapes and data graph would obviously not be
>         enough and we
>         > need to somehow introduce time, but not sure how.
>         >
>         > Best,
>         > Dimitris
>         >
>         > On Sat, Oct 1, 2016 at 3:25 AM, Peter F. Patel-Schneider
>         > <pfpschneider@gmail.com <mailto:pfpschneider@gmail.com>
>         <mailto:pfpschneider@gmail.com <mailto:pfpschneider@gmail.com>>> wrote:
>         >
>         >     What happens if a SPARQL query in SHACL is non-deterministic? 
>         Is there any
>         >     guarantee about when and how often SPARQL queries are processed
>         to produce a
>         >     query result?
>         >
>         >     For example are there any guarantees that
>         >
>         >     **prefixes, etc., as needed**
>         >
>         >     s:s1 rdf:type sh:Shape ;
>         >      sh:targetClass ex:c1 ;
>         >      sh:property [ sh:predicate ex:p1 ;
>         >                    sh:qualifiedMinCount 1 ; sh:qualifiedMaxCount 1 ;
>         >                    sh:qualifiedValueShape [
>         >                      sh:property [ sh:predicate ex:p2 ;
>         >                                    sh:shape s:s2 ] ] ] .
>         >
>         >     s:s2 rdf:type sh:Shape ;
>         >      sh:sparql [ sh:prefixes ex: ;
>         >       sh:select """SELECT $this WHERE {
>         >                  BIND ( ( RAND() * 25 )  AS ?r )
>         >                  FILTER ( STRLEN($this) < ?r ) }""" ] .
>         >
>         >     will never produce a violation on the data graph
>         >
>         >     PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#
>         <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
>         >     <http://www.w3.org/1999/02/22-rdf-syntax-ns#
>         <http://www.w3.org/1999/02/22-rdf-syntax-ns#>>>
>         >     PREFIX ex: <http://ex.org/>
>         >
>         >     ex:i1 rdf:type ex:c1 ;
>         >      ex:p1 [ ex:p2 ex:i3 ], [ ex:p2 ex:i3 ] .
>         >
>         >
>         >     Peter F. Patel-Schneider
>         >     Nuance Communications
>         >
>         >
>         >
>         >
>         > --
>         > Dimitris Kontokostas
>         > Department of Computer Science, University of Leipzig & DBpedia
>         Association
>         > Projects: http://dbpedia.org, http://rdfunit.aksw.org,
>         http://aligned-project.eu
>         > Homepage: http://aksw.org/DimitrisKontokostas
>         <http://aksw.org/DimitrisKontokostas>
>         > Research Group: AKSW/KILT http://aksw.org/Groups/KILT
>         >
> 
> 
> 
> 
> -- 
> Dimitris Kontokostas
> Department of Computer Science, University of Leipzig & DBpedia Association
> Projects: http://dbpedia.org, http://rdfunit.aksw.org, http://aligned-project.eu
> Homepage: http://aksw.org/DimitrisKontokostas
> Research Group: AKSW/KILT http://aksw.org/Groups/KILT
>

Received on Sunday, 2 October 2016 23:36:39 UTC