AW: AW: AW: {Disarmed} Re: blank nodes (once again)

Sandro Hawke wrote:

>On Thu, 2011-03-24 at 16:07 +0000, Michael Schneider wrote:
>> Hi Sandro!
>>
>> Sandro Hawke wrote:
>>
>> > Is there something in the OWL specs that says OWL doesn't work
>> > (or that we're no longer in DL) if the nodes composing the lists
>> > are not blank? That would be a problem.
>>
>> In the OWL 2 Mapping to RDF Graphs at
>>
>>     <http://www.w3.org/TR/2009/REC-owl2-mapping-to-rdf-20091027/>
>>
>> many of the RDF graph patterns on the left-hand-side of inverse mapping rules from RDF to >the Functional Syntax are only defined for blank nodes. See, for instance, Table 16, where you >can see the differences: there are some rules that use a string "*:x", which stands for an IRI >according to the definition in Chap 1, some rules which can be used with both IRIs and blank >nodes, as denoted by the string "x", and some of the rules in that table are really only defined >for blank nodes only, denoted by string "_:x". You can find many more examples for blank >node-only rules in the other tables.
>>
>> Specifically for lists, their inverse mapping is treated by Table 3, which only handles the case >of blank nodes.
>>
>> OWL 2 DL parsers may, of course, decide to relax on this and also parse structures having >URIs instead of blank nodes. But, strictly speaking, such RDF graphs do not count as valid OWL >2 DL ontologies in RDF graph form.
>
>Well, that's annoying.   Do you know of a technical reason for doing it
>this way, or was it just that no one saw a reason to allow nonblank
>nodes?

Look at the mapping rules, e.g. the one for intersection class expressions in Table 13 (first entry) of

    http://www.w3.org/TR/2009/REC-owl2-mapping-to-rdf-20091027/

The mapping result is the Functional Syntax expression

    ObjectIntersectionOf( CE(y1) ... CE(yn) )

There is nothing in the Functional Syntax that corresponds to the blank node of the RDF representation. The blank node is purely an artifact of the RDF representation. If you would use a URI instead of a blank node, and provided the OWL parser being used allows this at all, then the URI would simply get lost in the mapping process, since it cannot be "stored" anywhere in the corresponding Functional Syntax expression or the whole ontology. Even if there would be a mapping rule allowing for URIs, when you do a round-trip from RDF to Functional Syntax and back to RDF, you will in any case receive the corresponding expression having a blank node, since the URI gets lost in the mapping from RDF to Functional Syntax, and since the forward mapping from Functional Syntax to RDF does not "invent" URIs (and if it would, then the produced URI would most probably be different from the original URI). 

Note that for OWL 2 DL, the Functional Syntax (or, more precisely, the Structural Specification) is the specifying syntax, not the RDF mapping, which is just some under many serializations. And in the Functional Syntax, the only thing that really corresponds to a blank node is an anonymous individual. "Glue nodes" of multi-triple representations have no correspondence in the Functional Syntax.

Cheers,
Michael

--
Dipl.-Inform. Michael Schneider
Research Scientist, 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
Stiftung Az: 14-0563.1 Regierungspräsidium Karlsruhe
Vorstand: Dipl. Wi.-Ing. Michael Flor, Prof. Dr. rer. nat. Ralf Reussner,
Prof. Dr. rer. nat. Dr. h.c. Wolffried Stucky, Prof. Dr. rer. nat. Rudi Studer
Vorsitzender des Kuratoriums: Ministerialdirigent Günther Leßnerkraus
==============================================================================

Received on Thursday, 24 March 2011 21:23:31 UTC