- From: Ivan Herman <ivan@w3.org>
- Date: Sat, 10 Dec 2011 18:06:04 +0100
- To: Gregg Kellogg <gregg@kellogg-assoc.com>
- Cc: RDF Web Applications Working Group WG <public-rdfa-wg@w3.org>
Gregg, First of all, ouch. You are absolutely right, I have made a mistake. However, trying to analyze why I made that mistake I realized that what I did was to think in semantics and not in explicit rules, ie, I simply looked at the OWL RL rules instead of the old set of RDFS rules. Indeed, looking at the semantics of these rules: the missing rdfs5 and rdfs11 (more exactly, their equivalents in OWL RL) are _not_ necessary for the goal of vocabulary expansion. Indeed, these rules add extra knowledge about the properties and classes that appear in the vocabulary, but they do not lead tp any new extra triples that could be derived for the instance data (and could not be derived without them)! And, after all, the goal of the vocab expansion is to provide the minimal set of rules or, if you like, minimal semantics that is necessary to derive additional triples on the facts, on the instance data. So... in my view, what _is_ in the current draft is enough and the previous version was, in fact, slightly overspecified in terms of rules. A but in the previous version, if you like. It was not wrong, it was simply generating extra triples that were unnecessary for what was at hand. Note that, in case we want to reproduce the previous status with the new owl terms, then we need more than just two rules, because there are equivalent rules that bind the property equivalence with sub properties. We would need to add rules like scm-eqc2: ?c1, rdfs:subClassOf, ?c2 ; ?c2, rdfs:subClassOf, ?c1 => ?c1, owl:equivalentClass, ?c2 etc. Just for the records, here are the identifiers of the OWL 2 RL rules[1] that should be listed in case we want to add those: scm-sco, scm-eqc1, scm-eqc2, scm-spo, scm-eqp1, scm-eqp2 where scm-sco and scm-spo are the equivalents of rdfs11 and rdfs5, respectively, the other four bind the subclass and subproperty to equivalentclass and equivalentproperty, respectively. My proposal would be to leave the draft as is, and I have to apologize not to realize that when I did the change! Ivan [1] http://www.w3.org/TR/owl2-profiles/#Reasoning_in_OWL_2_RL_and_RDF_Graphs_using_Rules On Dec 9, 2011, at 03:43 , Gregg Kellogg wrote: > I'm just going through my code relative to vocabulary expansion. We agreed to add owl:equivalentProperty and owl:equivalentClass, but in transforming from RDFS to OWL, it looks like we lost some rules, is this intentional? > > Was: > > rdfs5: uuu rdf:subPropertyOf vvv . vvv rdfs:subPropertyOf xxx . => uuu rdfs:subPropertyOf xxx > rdfs7: aaa rdfs:subPropertyOf bbb . uuu aaa yyy . => uuu bbb yyy > rdfs9: uuu rdfs:subPropertyOf xxx . vvv rdf:type uuu . => vvv rdf:type xxx > rdfs11: uuu rdf:subClassOf vvv . vvv rdfs:subClassOf xxx . => uuu rdfs:subClassOf xxx > > Now: > > prp-spo1: p1 rdfs:subProperty p2 . x p1 y . => x p2 y (same as rdfs:7) > prp-eqp1: p1 owl:equivalentProperty p2 . x p1 y . => x p2 y (new, like rdfs:7) > prp-eqp2: p1 owl:equivalentProperty p2 . x p2 y . => x p1 y (inverse of prp-eqp1) > cax-sco: c1 rdfs:subClassOf c2 . x rdf:type c1 . => x rdf:type c2 (like rdfs9) > cax-eqc1: c1 owl:equivalentClass c2 . x rdf:type c1 . => x rdf:type c2 (new, like rdfs9) > cax-eqc2: c1 owl:equivalentClass c2 . x rdf:type c2 . => x rdf:type c1 (inverse of caxeqc1) > > It seems that we've lost equivalents for rdfs7 and rdfs11 along the way. I don't recall that this was intentional, but perhaps I missed something. It seems to me, we should add the following rules to the document: > > scm-spo: p1 rdfs:subPropertyOf p2 . p2 rdfs:subPropertyOf p3 . => p1 rdfs:subPropertyOf p3 (like rdfs5) > scm-sco: c1 rdfs:subClassOf c2 . c2 rdfs:subClassOf c3 . => c1 rdfs:subClassOf c3 (like rdfs11) > > Gregg > > On Nov 15, 2011, at 9:36 AM, RDF Web Applications Working Group Issue Tracker wrote: > >> >> ISSUE-116 (vocab-owl): Consider owl terms for vocab expansion [2nd LC Comment - RDFa Core 1.1] >> >> http://www.w3.org/2010/02/rdfa/track/issues/116 >> >> Raised by: Ivan Herman >> On product: 2nd LC Comment - RDFa Core 1.1 >> >> Add owl:equivalentClass and owl:equivalentProperty to the allowed terms for vocabulary expansion >> >> >> See the mail of Dan Brickley: http://www.w3.org/mid/CAFNgM+akmboWOgi5vSchetZHUFeiwORHDAHazR0DZAcCCnj33w@mail.gmail.com >> >> Relevant responses: >> http://www.w3.org/mid/2BF38B67-9AB6-49DC-9AFD-EEDCF3E15C92@w3.org >> http://www.w3.org/mid/4EC14415.9060609@aptest.com >> http://www.w3.org/mid/DEDB6120-6BEE-47CE-AE83-B9CF8D52D849@greggkellogg.net >> >> >> > > ---- Ivan Herman, W3C Semantic Web Activity Lead Home: http://www.w3.org/People/Ivan/ mobile: +31-641044153 FOAF: http://www.ivan-herman.net/foaf.rdf
Received on Saturday, 10 December 2011 17:08:54 UTC