Search Mailing List Archives
[p4-feedback] Problems with data properties and disjointness
tredmond at stanford.edu
Wed Apr 22 16:40:05 PDT 2009
I can add a little information to this. I suspect that part of the
problem here is that the OWL 1.1/2.0 -> RDF mapping has been a bit of
a moving target. In owl 1.1 , the statement that properties dpi
and dpj are disjoint is represented as
T(dpi) owl11:disjointDataProperties T(dpj)
In owl 2.0 , you have two choices of syntax. Either
T(DPE1) owl:propertyDisjointWith T(DPE2)
_:x rdf:type owl:AllDisjointProperties .
_:x owl:members T(SEQ DPE1 ... DPEn)
It looks like protege 4 got stuck in the middle of this transition
somehow. It does read and write the second version of the owl 2.0
format correctly. It also writes a version of the pairwise
disjointness property that is somewhere between the owl 1.1 version
and the owl 2.0 version. But as far as I can tell it cannot read
either version of the pairwise disjointness property.
On Apr 22, 2009, at 2:34 AM, S Essaid wrote:
> Several observations:
> 1. Disjoint declarations in RDF/XML not being recognized when file
> is opened. I attached an example file (DisjointNotReadIn.owl) and P4
> will not show the disjoint declaration in this OWL file. This is not
> a problem if OWL/XML is used.
> 2. If I declare three data properties ("a", "b", "c"), select "a",
> click disjoint icon, select "b" and "c" (in the same window, not one
> at a time), then all three are declared all different. This is not
> what I meant, "b" and "c" should not be disjoint from each other. It
> seams that selecting more than one property in the disjoint
> selection window is interpreted as indicating that all properties
> involved should be declared all disjoint.
> 3. When I tried to correct the error in 2 by selecting "b" and
> removing "c" as a disjoint (by clicking on "a" from the disjoint
> selection window that started with both "a" and "c" highlighted),
> then "c" lost its declaration as being disjoint from "a". Then if I
> select "c" in the list of properties and try to add "a" as a
> disjoint property, nothing happens. If I go to "a" and try to add
> "c" as a disjoint, again nothing happens.
> 4. Here are files for each of the following steps. File
> correctingalldisjoint.1.owl is after issue 2 above. File
> correctingalldisjoint.2.owl is after issue 3 above.
> 5. Another interesting situation can be reproduced like this (I was
> using OWL/XML):
> Create aa, bb, cc as data properties.
> Select aa and click disjoint properties.
> Select both bb and cc and click ok. (all disjoint will created for
> aa, bb, cc)
> Select bb and double click the disjoint statement, both aa and cc
> will be highlighted, click on aa and then ok. The all disjoint is
> now removed (looking at the RDF/XML)
> Select cc and click disjoint properties, aa is highlighted, click aa
> again and then ok. Nothing happens.
> Select aa and click disjoint properties icon. Click cc and then ok.
> Nothing happens to aa. Look at bb, cc was add as a disjoint to bb
> instead of cc being added to aa.
> After playing around by selecting one of the properties and adding
> the other as a disjoint ( I think it was after I selected bb and
> removed cc as a disjoint and then came back to bb and tried to add
> cc again as a disjoint) I got the RDF/XML shown in the attached
> interesting.owl file.
> 6 Some of these issues were tried using object properties and I
> noticed some of the same problems.
> 7 Everything worked as expected if I do not select more than one
> property as a disjoint from the selection window. But doing this
> will not create all disjoint statements in the RDF/XML even if they
> are appropriate. It seams that applying disjoint properties does not
> work correctly once an all disjoint statement is created by
> selecting more than one disjoint property from the selection window.
> Even when all the disjoints are later removed from all the
> properties (confirmed by looking at the RDF/XML) the user interface
> still does not behave correctly. The ontology has to be closed and
> reopened to fix this issue. The lonely all different declaration can
> be removed by removing the empty disjoint entry (empty in the GUI)
> for that property.
> 8 Another strange issue is shown in the attached screenshot. I was
> trying to add dt:bb as a disjoint property to dt:dd. You can see
> dt:dd selected on the left and on the right is the window for
> selecting the disjoint property. Notice that dt:aa and its
> subproperties are not visible in the selection window.
> p4-feedback mailing list
> p4-feedback at lists.stanford.edu
More information about the p4-feedback