RE: error in RDF TR on rdf:predicate

>-----Original Message-----
>From: public-owl-wg-request@w3.org [mailto:public-owl-wg-request@w3.org]
>On Behalf Of Peter F. Patel-Schneider
>Sent: Wednesday, May 13, 2009 5:19 AM
>To: public-owl-wg@w3.org
>Subject: error in RDF TR on rdf:predicate
>
>I believe that I have found an error in http://www.w3.org/TR/rdf-schema/
>at http://www.w3.org/TR/rdf-schema/#ch_predicate (even though I suppose
>that it actually is permissable in RDF to be in this strange state).
>
>--------------------------
>
>5.3.3 rdf:predicate
>
>rdf:predicate is an instance of rdf:Property that is used to state the
>predicate of a statement.
>
>A triple of the form:
>
>    S rdf:predicate P
>
>states that S is an instance of rdf:Statement, that P is an instance of
>rdf:Property and that the predicate of S is P.
>^^should be rdf:Resource
>
>The rdfs:domain of rdf:predicate is rdf:Statement and the rdfs:range is
>rdfs:Resource.
>
>--------------------------
>
>The strange thing is that this was explicitly changed to conform with
>the
>RDF Semantics, but the change wasn't fully carried through.  Even
>stranger, it appears that the change probably should have been made the
>other way!
>
>peter
>
>PS:  Of course, the strangest part of all this is that I found this when
>     researching a response to one of Jeremy Carroll's 666 lmLCCs.
>
>PPS:  Maybe even stranger is the fact that a similar issue crept into
>      the OWL 2 RDF-Base Semantics.

A few days ago, when writing testcases for the OWL Full treatment of annotations, I found what I considered a stupid bug of myself and fixed it: The range of owl:predicate was defined to be rdf:Property, but should IMO be rdfs:Resource. The rational is that one of my design principles has always been that annotations in OWL Full, while obviously not semantic free, should at least not add unnecessarily to the semantics compared to what the RDF semantics already provides. 

Now if I have a triple

  _:x owl:predicate ex:p

what the RDFS Semantics tells me is that ex:p denotes a resource, but nothing more. So not more should be said about ex:p under the OWL 2 RDF-Based Semantics, either. 

To summarize, I consider this a simple bug w.r.t. a design principle that I followed everywhere else in the spec. There is no change of design, rather the old version was incompatible with the existing design, hence a bug. I have added both a comment on this in the text preceding the semantic conditions on properties (§5.3), and in the list of Post-LC changes. I also plan to add a sentence to the Introduction (to the paragraph talking about annotations) along the lines I stated above.

Concerning the RDF Reification vocabulary: I guess the situation is meant differently there. There, the idea (though not normatively stated) was something like that a reified triple is about the symbols in the RDF graph, rather than about the resources denoted by these symbols. If one buys this (I do not), then it's somewhat clear that the range of rdf:predicate can't be rdf:Property, since such symbol will generally not be a property, although it will be at least have to be an instance of the domain of discourse, i.e. a resource. But even if one (as I do) reads a reified triple as an assertion about the relationship between the /denoted resources/, I would still not like to see the range of rdf:predicate to be rdf:Property, since this may say more about the value of a rdf:predicate triple than one is willing to state. But rdf:Reification is luckily off-topic for us -- and hopefully keeps so. :-)

Michael

--
Dipl.-Inform. Michael Schneider
Research Scientist, Dept. Information Process Engineering (IPE)
Tel  : +49-721-9654-726
Fax  : +49-721-9654-727
Email: michael.schneider@fzi.de
WWW  : http://www.fzi.de/michael.schneider
=======================================================================
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, RP Karlsruhe
Vorstand: Prof. Dr.-Ing. Rüdiger Dillmann, Dipl. Wi.-Ing. Michael Flor,
Prof. Dr. Dr. h.c. Wolffried Stucky, Prof. Dr. Rudi Studer
Vorsitzender des Kuratoriums: Ministerialdirigent Günther Leßnerkraus
=======================================================================