On 2021-09-07 3:07 pm, Scott Henninger wrote:
Thanks for the help Holger, it gave me some ideas, which unfortunately
haven't panned out yet. First, I take it that the parent in the
third-to-last line of the example should be a square bracket. Or
there's something else missing in the syntax.
Right, it was lacking a sh:property [ ... ] around the last two rows.
It's hard to help on such issues without real test data, so moving
forward it would be best to get real instances, real class definitions
and a description of invalid instances so that we only need to fill in
the blanks. I see you have started this below.
The most confounding part to me is how to express the class
constraints, so let me focus there with three example data sets.
This is a valid shape. The target class has both Cls1 and Cls2 and
the parent class has one of Cls{3, 4, 5}:
ex:target1
a ex:Cls1, ex:Cls2 ;
ex:parent ex:target2 .
ex:target2
a ex:Cls3 .
This is an invalid shape only because of the extra class definition in
target1. The extra class definition in target2 is OK.
ex:target1
a ex:Cls1, ex:Cls2, ex:Cls6;
ex:parent ex:target2 .
ex:target2
a ex:Cls3, Cls7 .
This case could become
ex:Shape
a sh:NodeShape ;
sh:targetClass ex:Cls1, ex:Cls2 ;
sh:not [ sh:class ex:Cls6 ] ;
to state that instances of Cls1 or Cls2 cannot also be instances of
Cls6. It does NOT state that they need to have both types 1 and 2.
This is also an invalid shape because target2 does not have a class
definition in Cls{3, 4, 5}:
ex:target1
a ex:Cls1, ex:Cls2;
ex:parent ex:target2 .
ex:target2
a ex:Cls7 .
This case could become
ex:Shape
a sh:NodeShape ;
sh:targetClass ex:Cls1, ex:Cls2 ;
sh:property [
sh:path ex:parent ;
sh:or (
[ sh:class ex:Cls3 ]
[ sh:class ex:Cls4 ]
[ sh:class ex:Cls5 ]
)
]
But this logic feels different from what you had in your original SPARQL
query, which seemed to be more "exists" than "for-all".
Note that for general non-product related questions you may get better
breadth of input on places like StackOverflow or the
https://www.w3.org/community/shacl/ mailing list or similar places. This
is not to discourage you from asking here, but to understand that we
cannot give such requests the same amount of attention as customer
questions, so my input may be rather cursory than in-depth.
Thanks,
Holger
Hopefully this helps some. As arbitrary as it seems, the examples
come from real requirements, and while I'm pretty familiar with
targeting a class, in this case two classes are being targeted along
with a property-to-type relationship. That I'm unclear on, so any
hints on any pieces of this puzzle would be useful.
Thanks again for your efforts!
-- Scott
On Sun, Sep 5, 2021 at 4:15 PM Holger Knublauch
<[email protected] <mailto:[email protected]>> wrote:
On 2021-09-05 3:38 am, Scott Henninger wrote:
Unfortunately we do not have SPARQL support for SHACL, as we use
an old version of Semaphore. Maybe someday in the future.
In the meantime, what about if we drop the sub-property triple
pattern and have a known set of (or single) properties? Can that
be expressed in SHACL? Particularly the part about targeting
multiple classes and multiple parent classes.
Ok, if the properties are known you can use a pattern such as
sh:path [ sh:alternativePath ( ex:prop1 ex:prop2 ex:prop3 ) ]
; # ex:prop1|ex:prop2|ex:prop3
The rest could roughly be
ex:Shape
a sh:NodeShape ;
sh:targetClass ex:Cls1 ;
sh:not [
sh:class ex:Cls2 ;
sh:property [
sh:path ex:parent ;
sh:qualifiedValueShape [
sh:or ( [ sh:class ex:Cls3 ] [ sh:class ex:Cls4 ]
[ sh:class ex:Cls5 ] ) ;
] ;
sh:qualifiedMinCount 1 ;
) ;
sh:path [ sh:alternativePath ( ex:prop1 ex:prop2 ex:prop3
) ] ; # ex:prop1|ex:prop2|ex:prop3
sh:minCount 1 ;
]
(above I assume your rdf:type matches can be expressed using
sh:class, i.e. subclasses also count).
I am not sure if the above represents your intention, but the
mechanism that I have used was basically to match the pattern that
you have in the WHERE clause and put a sh:not around it, meaning
that a violation will be reported if the instance of Cls1 also
matches the given pattern.
Holger
Thanks, Holger! I appreciate the help.
-- Scott
On Sep 3, 2021, at 8:06 PM, Holger Knublauch
<[email protected]> <mailto:[email protected]> wrote:
Hi Scott,
I may not fully understand your scenario, but I don't think this
can be expressed in SHACL Core. The problem is not so much in
the matching pattern (with the types etc) but in the
subPropertyOf link. The last line would need to become something
with sh:qualifiedMinCount 1 but you would need a sh:path and the
sh:path cannot be computed dynamically, i.e. you cannot use a
variable that gets computed using subPropertyOf. It would only
work if you can know the sub-properties in advance, when you
write the constraint.
You have it formulated in SPARQL already, so why not stay in
SHACL-SPARQL? Does your target platform only support SHACL Core?
Holger
On 2021-09-03 4:01 pm, Scott Henninger wrote:
Another SHACL question for the topbraid-users group:
This one is a bit hard to explain, but the general idea is to
create a shape that requires two target classes, and the
hierarchical parent must be one of three classes, and the
target cannot have any properties are a sub-property of a property.
Working it out a bit I believe the following is the equivalent
query in SPARQL:
SELECT *
WHERE {
$this a ex:Cls1, ex:Cls2 .
{ $this ex:parent/rdf:type ex:Cls3 . }
UNION
{ $this ex:parent/rdf:type ex:Cls4 . }
UNION
{ $this ex:parent/rdf:type ex:Cls5. }
?prop rdfs:subPropertyOf ex:assoc .
$this ?prop ?x
}
Any $this found by this query would be a violation, based on
the last triple pattern - i.e. the rest are requirements,that
are taken up by other shapes. But what I want here is to
report if the target class combination, with appropriate
parents, has one of the ex:assoc sub-properties. To
paraphrase: "Instances of Cls1 and Cls2 ... cannot have any
ex:assoc properties.
Any hints or direction on this would be useful. I've tried a
few things but have so far failed to make much headway.
Appreciate it and TIA.
-- Scott
--
You received this message because you are subscribed to the
Google Groups "TopBraid Suite Users" group.
To unsubscribe from this group and stop receiving emails from
it, send an email to
[email protected]
<mailto:[email protected]>.
To view this discussion on the web visit
https://groups.google.com/d/msgid/topbraid-users/3a55f966-d99d-4d32-875e-3e32a1f643c4n%40googlegroups.com
<https://groups.google.com/d/msgid/topbraid-users/3a55f966-d99d-4d32-875e-3e32a1f643c4n%40googlegroups.com?utm_medium=email&utm_source=footer>.
--
You received this message because you are subscribed to the
Google Groups "TopBraid Suite Users" group.
To unsubscribe from this group and stop receiving emails from
it, send an email to [email protected]
<mailto:[email protected]>.
To view this discussion on the web visit
https://groups.google.com/d/msgid/topbraid-users/cca816bc-26db-4af4-2881-4ee80ff7866e%40topquadrant.com
<https://groups.google.com/d/msgid/topbraid-users/cca816bc-26db-4af4-2881-4ee80ff7866e%40topquadrant.com?utm_medium=email&utm_source=footer>.
--
You received this message because you are subscribed to the
Google Groups "TopBraid Suite Users" group.
To unsubscribe from this group and stop receiving emails from it,
send an email to [email protected]
<mailto:[email protected]>.
To view this discussion on the web visit
https://groups.google.com/d/msgid/topbraid-users/49677E3C-2B4F-488F-BB6B-90881B496020%40gmail.com
<https://groups.google.com/d/msgid/topbraid-users/49677E3C-2B4F-488F-BB6B-90881B496020%40gmail.com?utm_medium=email&utm_source=footer>.
--
You received this message because you are subscribed to the Google
Groups "TopBraid Suite Users" group.
To unsubscribe from this group and stop receiving emails from it,
send an email to [email protected]
<mailto:[email protected]>.
To view this discussion on the web visit
https://groups.google.com/d/msgid/topbraid-users/baad61dd-3f49-84e6-09d1-8db48331c1af%40topquadrant.com
<https://groups.google.com/d/msgid/topbraid-users/baad61dd-3f49-84e6-09d1-8db48331c1af%40topquadrant.com?utm_medium=email&utm_source=footer>.
--
You received this message because you are subscribed to the Google
Groups "TopBraid Suite Users" group.
To unsubscribe from this group and stop receiving emails from it, send
an email to [email protected]
<mailto:[email protected]>.
To view this discussion on the web visit
https://groups.google.com/d/msgid/topbraid-users/CAJ-KvVqCG_3HZUDM-NL_PR4i%2B6c4RQfL%3DyNjR5Odco5exs%3DJMg%40mail.gmail.com
<https://groups.google.com/d/msgid/topbraid-users/CAJ-KvVqCG_3HZUDM-NL_PR4i%2B6c4RQfL%3DyNjR5Odco5exs%3DJMg%40mail.gmail.com?utm_medium=email&utm_source=footer>.
--
You received this message because you are subscribed to the Google Groups "TopBraid
Suite Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/topbraid-users/4d9e4f02-405f-f0c5-7969-ff2cf1e3d11a%40topquadrant.com.
