Re: [OWL-S] DLization of 1.1

Bijan,

Just a couple of small points on your new List:

1) the rest property should have List as its range as well as domain.

2) OWL DL doesn't allow renaming built-in rdf, rdfs or owl classes or 
properties[1], so they have to be called something other than 
List/first/rest.

Is this list construct going to be used in OWL-S or is the issue not 
settled?

Regards,
/Daniel


[1] http://www.w3.org/TR/2004/REC-owl-ref-20040210/ Appendix E, first point

Bijan Parsia wrote:

>
> Here's some notes from my DLization efforts. Thanks to those who went 
> before :) (e.g., Sean and Peter)
>
> The files are at:
>     http://www.mindswap.org/~bparsia/ontologies/sws/owls1.1/
>
> They will be migrated to daml.org soon.
>
> My comments on just about every change are inline in xml comments 
> starting with "BJP:"
>
> Methodology:
>     I worked out fixes on my own first, then compared with other like 
> minded efforts. (Still have to double check         some)
>     I checked species with Pellet and with the WonderWeb species 
> validator.
>     I did also try to do minor improvements as I went along (e.g., 
> adding xml:bases).
>     I tried to test for consistency, unsatisfiable concepts, etc. but 
> the pellet web service kept timing out on me
>         for grounding. I'll follow up eventually.
>
> Things that would be nice:
>     To see if the Protege OWL plugin can load these.
>     Test with various reasoners and other editors.
>
>
> time-entry.owl:
>     seems to have gotten an xml:base, yay!
>
> Service.owl:
>     Use of rdfs:Resource in a domain. I just eliminated the domain 
> claim as it's implied away.
>     
>     I would like to remove the redundant global restrictions on the 
> inverse property, but I recongnize that it might     be helpful to 
> have them explicit for dumber tools and for human readers. Any strong 
> thoughts one way or the
>     other?
>
>     Added an xml:base
>
> Process.owl:
>     Oy!
>     Added xml:base
>     Changed the &xsd; entity declaration to include the #. Changed the 
> one ref (&xsd;#boolean) to match
>     Added a &swrl; entity declaration. Are we going to use the SWRL 
> ontology?
>     Added type owl:Ontology to the import.
>     Replace parameterType with new DatatypeProperty as indicated in my 
> prior email.
>     Cleaned up Process cardinality claim.
>     hasParticipant is now an ObjectProperty (why is this a 
> subPropertyOf hasParameter??)
>     computedInput, computedPreconditoin, computedOutput, et al are now 
> an ObjectProperty as their ranges         indicates (will these end up 
> like Conditions?, i.e,. formulas?) The range is &owl;#Thing, which, IMHO
>         should be surpressed as it's redundant *and* we're likely to 
> want to replace it with something
>         appropriate.
>     I've added a minimal "Shadow list" vocabulary, to handle process 
> componant bag, et al. This isn't
>         complete, in the senses that 1) I've not worked it up into a 
> full ontology and 2) I've not made it
>         useful for defining parameterTypes, and 3) I've not done 
> anything toward parseType=Collection
>         sugar.
>     Binding was missing appropriate property nodes. I added 
> subClassOfs. Should be checked with
>         author to see if that expresses intentions. Actually, I'm not 
> sure what's up here. BInding has
>         exaclty one formal and actual parameter? Don't quite get it.
>           Input and Output Bindings needed subClassOfs.  Simimlar 
> issues with OutputBinding.
>
> The W3C RDF validator is still better than pellet or wonderweb at 
> catching these rdfy problems.
>
>     Bunch more rdf:Property to ObjectProperties.
>
> Grounding.owl
>     Added a type Ontology to the import.
>     Changed the &xsd; entity to include the hash. Would like that to 
> be the style
>     Fixed a buncha cardinality bugs
>     Added objectproperty decls to all the functionalproperties
>         I adopted the idiom of always using the fundemental type 
> (Data|ObjectProperty) as the typed node
>         bearing the rdf:ID, and then add, outside the "definition" 
> various qualifying typed nodes, e.g.,:
>             <owl:ObjectProperty rdf:ID="owlsProcess">
>                   <rdfs:comment>
>                       .....
>             </owl:ObjectProperty>
>             <owl:FunctionalProperty rdf:about="#owlsProcess"/>
>         This seems more readable and editable.
>         I would like to convert the existing rdf:type FunctionalProp 
> to this style. The main annoyance is if you
>         change the name of the property. Either way, we should be 
> consistent.
>
> Profile.owl:
>
>     Added xml:base
>     Upgraded &xsd;
>     Fixed cardinality
>     Fixed missing object properties, etc.
>
> *****
> Bugs in the validators:
>
> Both Pellet and wonderweb pass:
>       <owl:Restriction owl:cardinality="1">
>           <owl:onProperty rdf:resource="#owlsProcess"/>
>           </owl:Restriction>
>
> WonderWeb doesn't seem to do any datatype checking at all, even on 
> cardinalities.
>
> WonderWeb passes:
>       <owl:imports>
>             <owl:Ontology rdf:resource="&processImport;"/>
>       </owl:imports>
>
> (Should be rdf:about, or the like.)
>
> Cheers,
> Bijan Parsia.
>

Received on Wednesday, 7 April 2004 02:50:17 UTC