RE: ACTION-181: Review RDF-Based Semantics.

Alan Wu wrote:

>Hi,
>
>Sorry for this very late email. I have finished reviewing RDF-based
>semantics document. It is indeed very good work. Thanks Michael!

Many thanks, Zhe!

>My comments are embedded in the document.

Well, not anymore. :) See my comments below!

     Vocabulary:
     """
     I could not find reference of owl:datatypeArity 
     in Semantics, Syntax, or Mapping document.
     """

Yes, things have changed in those documents recently. I have updated the Full semantics accordingly.

     Definition of Consistency:
     """
     Is Q used in this definition?
     """

Thanks, this was a bug. Fixed.  

     Table 4.1: "Basic Sets"
     """
     A simple description of these names will be convenient.
     """

Agreed. I have added a column with explanations.

     Table 4.2: Convenient Abbreviations
     """
     A minor point, to be consistent, 
     shall we call it IF(d) and IFV(d,f) for the next one?
     """

Uh, a naming issue! :-] 

Then I suggest (now in the document this way):

  * IFACET --> IFA
  * IFACETVAL --> IFAV
  * IFEXT --> IFAEXT

Rational: That's a bit closer to the naming convention in the DL documents, which use "FA" as a base (V_FA, N_FA, ...). 

    Table 4.4: "Properties"
    """
    It appears that element in ILIST can be a list of anything. 
    Do we need to restrict the list to be a list of classes?
    """

Do you mean a kind of "generic": "ILIST<IC>"?

I agree that having a more specific range of "owl:disjointUnionOf" would be a good thing. But I'm afraid this would lead to some complications. Note that ILIST is simply an abbreviation for "ICEXT(IS(rdf:List))". The expression

  IEXT(IS(owl:disjointUnionOf)) subset IC x ILIST

can be translated into axiomatic triples having a very simple form:

  owl:disjointUnionOf rdfs:domain rdfs:Class
  owl:disjointUnionOf rdfs:range rdf:List

But, AFAICT, ILIST<IC> cannot be expressed that simple in triple form, at least not by existing means. One could introduce a new URI 

  owl:CList rdfs:subClassOf rdf:List ;
            owl:hasMembersOfClass <C> .

give appropriate semantics to it, and then use this new URI for the purpose of specifying the range of certain classes more precisely. But, I suppose, we probably won't go this way, and that's all very speculative at the moment. :)

Note that the RHS of the semantic condition of owl:disjointUnionOf tells that all the list members are classes, so the desired information exists at least when the language feature is used somewhere in an ontology. While this is not quite the same, the situation is not that bad. And having ILIST is at least still a good hint that this construct may be used with more than one argument, so it's better than just IR.

So I am not going to change this at the moment.

    Relationship to OWL 2 DL
    """
    Not sure if this is too much to ask. 
    The OWL 1 SAS document describes in Section 5 
    common part of OWL 1 DL semantics and OWL 1 FULL semantics. 
    Later, differences between them are introduced. 
    Can we have some high level description of commonality 
    between OWL 2 DL and OWL 2 FULL semantics? 
    Maybe it is hard because of the the fundamental difference described below?
    """

What you are referring to was not really the description-logic-based OWL 1 DL, but an RDFS compatible language (also called "OWL DL"), which happened to be entailment-equivalent (-> Theorem 1 in sec. 5.4) to the "real" OWL DL language for those RDF graphs, which are valid OWL DL ontologies in RDF graph form. There is no such language anymore in OWL 2 (that was the scope of ISSUE-121, which was not opened by the chairs), so I see no reason to treat this now non-existing language in the OWL 2 Full spec in any form.

(And if it would exist in OWL 2, I would rather opt to have it specified in a document separate from the OWL Full spec. From my side, I would /not/ be willing to spend any time on it.)

    Comprehension Principles
    """
    Please specify the scope of x here. For example, is x an element of IR or IC, etc.
    """

Ok, added. Although not technically necessary (if nothing is said, then it's unconstrained, i.e. it's IR), this helps perhaps better understanding the document. 

Regards,
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