- From: Guus Schreiber <schreiber@cs.vu.nl>
- Date: Thu, 28 Aug 2003 00:00:04 +0100
- To: Sean Bechhofer <seanb@cs.man.ac.uk>
- Cc: Jim Hendler <hendler@cs.umd.edu>, webont <www-webont-wg@w3.org>, Jeremy Carroll <jjc@hpl.hp.com>
Response wrt. point 2. Sean Bechhofer wrote: > On Tue, 26 Aug 2003, Jim Hendler wrote: >>>> 2) >>>> >>>> Test case: >>>> http://www.w3.org/2002/03owlt/miscellaneous/Manifest102 >>>> Input document: >>>> http://www.w3.org/2002/03owlt/miscellaneous/consistent102 >>>> >>>> The manifest declares the input as an OWL DL document. In section >>>>8.3 of the OWL Web Ontology Language Reference (regarding use of >>>>owl:intersectionOf in OWL Lite documents) it says: >>>> >>>> “owl:intersectionOf be used only on lists of length greater than >>>>one that contain only named classes and restrictions” >>>> >>>> This would appear to be the case for the input document, the >>>>object of the owl:intersectionOf triple is a list of length greater >>>>than one containing only restrictions. I think the input document >>>>is an OWL Lite document and the test case should be changed >>>>accordingly. > > > Section 4.1 of S&AS details the mapping rules that translate abstract > syntax to triples. Examination of the table shows that there are only > three ways in which an intersectionOf triple can occur: > > 1) Due to a class definition > > Class(classID [Deprecated] partial > annotation_1 ... annotation_m > description_1 ... description_n) > > 2) Due to an intersectionOf expression > > intersectionOf(description_1 ... description_n) > > 3) Due to a restriction with multiple components: > > restriction(ID component_1 ... component_n) > > Now 2) and 3) are only available in OWL DL (See Section 2.3 of S&AS). So > for an intersection to be valid in an OWL-Lite document, it must have come > from the translation of a class definition, and (from the right hand side > of the table in 4.1) it must therefore be the case that the subject of the > intersectionOf triple is a classID. This is *not* the case in > miscellaneous/consistent102, as the subject of the intersection is a blank > node. So the ontology is not Lite. > > I guess the upshot of this is that the restriction listed in section 8.3 > of Reference is not quite strong enough -- it is not only the case that > intersectionOf be used only on lists of length greater than one that > contain only named classes and restrictions -- in addition intersectionOf > cannot apply to blank nodes. Sec. 8.3 of Reference also contains the following constraint on OWL Lite (see just below the intersectionOf bullet): [[ * the object of rdf:type triples be named classes or restrictions; ]] so Ref is correct wrt. the test (102 has a blank node as object of rdf:type). Having said that, there could be room for improving the description of the Lite restrictions. However, I do not agree with Jeremy's proposed response: Jeremy Carroll wrote: http://lists.w3.org/Archives/Public/www-webont-wg/2003Aug/0122.html > 2) > http://www.w3.org/2002/03owlt/miscellaneous/consistent102 > > The file is in OWL DL not OWL Lite. > > An analysis of sections 2 and 4 of S&AS shows that in OWL > Lite there are no blank nodes of type owl:Class. The last statement is incorrect, as Lite allows (only) blank nodes of type owl:Restriction, a subclass of owl:Classs. Note also that the mapping rules for restrictions in Sec. 4.2 of S$AS have an optional "rdf:type owl:Class" triple. Therefore I propose to add the following statement to Ref, Sec. 83. [[ Blank nodes representing classes should be of type owl:Restriction. ]] and adapt the response accordingly. The addition is just for clarity: it is already implied by the constraints listed in 8.3. Guus > > Cheers, > > Sean > -- Free University Amsterdam, Computer Science De Boelelaan 1081a, 1081 HV Amsterdam, The Netherlands Tel: +31 20 444 7739/7718 E-mail: schreiber@cs.vu.nl Home page: http://www.cs.vu.nl/~guus/
Received on Wednesday, 27 August 2003 18:59:59 UTC