[OWLWG-COMMENT] ISSUE-67 (reification): REPORTED: use of reification in mapping rules is unwise

[
public comment to OWL-1.1-WG issue 67
cc'ed to WG-members involved in this discussion
]

Hi!

On Tue, 20 Nov 2007, Jeremy Carroll has raised the following issue:

  <http://lists.w3.org/Archives/Public/public-owl-wg/2007Nov/0369.html>

  ISSUE-67 (reification): 
  REPORTED: use of reification in mapping rules is unwise

  "The mapping rules use RDF reification.

  However, RDF reification has very weak semantics, 
  making it difficult to achieve an OWL Full semantics 
  that works.

  It is unlikely that  RDF graph's constructed with such rules mean what is
intended."


The wording in this issue is pretty generical, but, AFAICS, the ongoing
discussion in the WG about this issue is completely restricted to /axiom
annotations/.

I just want to bring to everyone's attention that axiom annotations are not
the only OWL-1.1 constructs which have an RDF mapping that uses RDF
reification. There are also so called "NegativePropertyAssertion"s, both for
object and data properties, which specify that a given "s p o" statement
does /not/ exist in the knowledge base, see Table 4 in [1].

As an example, let's regard the following ontology, given in functional
syntax:

  (A1) ObjectPropertyAssertion(p u w)
  (A2) NegativeObjectPropertyAssertion(p u w)

Now, if I map this ontology to RDF, according to Table 2 of [2] I receive
(ignoring the required additional typing triples):

  (R11) u p w .

  (R21) _:x rdf:type owl11:NegativeObjectPropertyAssertion .
  (R22) _:x rdf:subject u .
  (R23) _:x rdf:predicate p .
  (R24) _:x rdf:object w .

So the NegativeObjectProperty has been encoded by reifying the statement 'u
p w'.

The ontology {(A1),(A2)} is clearly contradictory according to the
OWL-1.1-DL semantics given in [1]. This would mean for a (yet-to-specify)
RDF-compatible OWL-1.1-Full semantics that a contradiction must be
entailable somehow from the RDF graph {(Rij)}. I think that the concerns
regarding the use of RDF reification, generically stated by Jeremy in
ISSUE-67 (see above), hold for NegativeObjectPropertyS, too. So I suggest
that ISSUE-67 should be attached a comment, which tells what OWL-1.1
constructs actually fall under ISSUE-67.

Otherwise the discussion in the WG might move into a wrong direction. For
example, Boris Motik discusses:

  <http://lists.w3.org/Archives/Public/public-owl-wg/2007Nov/0389.html>

  "There is yet another solution: we might have axiom annotations 
  in the structural specification, but then disallow (or simply 
  delete them) when the ontology is exported into OWL RDF."

While this would really be a technical solution for the case of
"semantic-less" annotations, an analog approach would obviously /not/ be a
working solution for the case of negative property assertions, because the
latter are real axioms, and therefore must be represented in every syntax.

Cheers,
Michael

[1] OWL-1.1 Semantics
    <http://www.w3.org/2007/OWL/wiki/Semantics>

[2] OWL-1.1 RDF-Mapping
    <http://www.w3.org/2007/OWL/wiki/Mapping_to_RDF_Graphs>

--
Dipl.-Inform. Michael Schneider
FZI Forschungszentrum Informatik Karlsruhe
Abtl. Information Process Engineering (IPE)
Tel  : +49-721-9654-726
Fax  : +49-721-9654-727
Email: Michael.Schneider@fzi.de
Web  : http://www.fzi.de/ipe/eng/mitarbeiter.php?id=555

FZI Forschungszentrum Informatik an der Universität Karlsruhe
Haid-und-Neu-Str. 10-14, D-76131 Karlsruhe
Tel.: +49-721-9654-0, Fax: +49-721-9654-959
Stiftung des bürgerlichen Rechts
Az: 14-0563.1 Regierungspräsidium Karlsruhe
Vorstand: Rüdiger Dillmann, Michael Flor, Jivka Ovtcharova, Rudi Studer
Vorsitzender des Kuratoriums: Ministerialdirigent Günther Leßnerkraus

Received on Thursday, 22 November 2007 23:14:17 UTC