W3C home > Mailing lists > Public > public-owl-wg@w3.org > March 2008

RE: ACTION-93 (Full semantics): Short report about current state

From: Michael Schneider <schneid@fzi.de>
Date: Wed, 12 Mar 2008 13:19:20 +0100
Message-ID: <0EF30CAA69519C4CB91D01481AEA06A08030F1@judith.fzi.de>
To: "Ivan Herman" <ivan@w3.org>
Cc: <public-owl-wg@w3.org>
Hi Ivan!

>Michael,
>
>thanks.
>
>I was just wondering... maybe it would be helpful to add a 
>more explicit 
>RDF syntax into the document. It may help the reader.
>
>As an example, I looked at the sub property chain[1]; I guess what you 
>describe there as a possible syntax can be translated into RDF/XML and 
>turtle as:
>
><rdf:Description>
>   <owl11:subObjectPropertyChain rdf:parseType="Collection">
>     <rdf:Description rdf:about="#p1"/>
>     <rdf:Description rdf:about="#p2"/>
>     ...
>     <rdf:Description rdf:about="#pn"/>
>   </owl11:subObjectPropertyChain>
>   <rdfs:subPropertyOf rdf:resource="#q"/>
></rdf:Description>
>
>[
>   owl:subObjectPropertyChain (<#p1> <#p2> ... <#p3>);
>   rdfs:subPropertyOf <#q>
>
>which looks pretty intuitive to me syntax-wise...

The presentation style for syntax and semantics in the Wiki will most probably not be what you will find later in the Editor's draft. The Wiki is for development purposes and for WG-internal review only.

Regarding the syntax: I usually take the syntax from the RDF mapping document and "clean it up" a bit. It is there just for convenience, mostly for me, but also for other WG members, who are willing to review. I prefer to use a syntax which explicitely enumerates the different triples, because this makes it easier to think about the respective semantics. I regard RDF/XML to be not a very good syntax for this purpose. And Turtle tends to hide information, in particular bNodes, in particular if more than one bNode is around.

Note, btw., that in the OWL-1.0-Full spec, the syntax isn't presented at all. 

>By the way, on the very same page, you raise the possibility for an 
>alternative syntax, namely:
>
>q owl11:subObjectPropertyChain ( p_1 ... p_n )
>
>ie, in Turtle:
>
><#q> owl11:subObjectPropertyChain (<#p1> <#p2> ... <#pn>) .
>
>At first glance this looks more compact but it might be 
>counter-intuitive. Indeed, we already have, from RDFS
>
><#a> rdfs:subPropertyOf <#b>
>
>In the latter, <#a> is the 'sub' and <#b> is the 'super'. On the other 
>hand, in the syntax above, <#q> is the 'super' and, well, the chain is 
>the 'sub':-( This may lead to errors...

Please wait! The sub property chains are still work in progress (marked as "WIP"), so they are not intended for review by the WG yet. Only categories which I have marked as "finished" in the table at

  <http://www.w3.org/2007/OWL/wiki/FullSemantics>

are meant to be ready for review. 

Anyway, because this proposal probably won't change /conceptually/ (the name of the property may change, of course): I have found problems with the current RDF encoding of sub property chains. For example, this encoding will produce certain undesired semantical artifacts in OWL-Full, which has to do with the fact that the URI 'rdfs:subPropertyOf' is reused for the purpose of defining sub property chains (the LHS of the subPropertyOf triple does *not* necessarily denote the property chain). There are other points which seem to speak against the current RDF syntax, but I am not finished with  investigating. 

For the question whether the list should be on the LHS or on the RHS: There is no precedence in OWL/RDF syntax with sequences on the LHS, so I would not like to introduce this if it is avoidable. I remember that the original RDF syntax for sub property chains had the sequence at the LHS. I experienced that this makes it somewhat hard to work with Topbraid Composer, where you normaly have some resource (e.g. some named property), and add a property together with a value. This paradigma does not directly support sequences on the LHS, so you have to trick around: First build a sequence and give it a name, and then add a subProperty property to this list resource.

Just a few preliminary thoughts on this.

>
>Ivan
>
>[1] http://www.w3.org/2007/OWL/wiki/FullSemanticsSubPropertyChains

Cheers,
Michael

--
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 Wednesday, 12 March 2008 12:19:50 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 12 March 2008 12:19:51 GMT