Re: foaf/OWL

On Nov 25, 2007, at 11:37 AM, Jim Hendler wrote:

> Alan,
>   Forgive me for disagreeing

With the ifdt, ok. I said I wasn't holding my breath. What about the  
other stuff?

With respect to that issue, there are a couple of things. First, I  
think the chatid properties need to be reworked. They need to be  
extensible - there ought to be a way to add a new chat service, an id  
in that service, and to be able to look up some information about how  
to connect. For example, there is no standard way to add a skype id  
currently.

That leaves only the foaf:mbox_sha1sum. Currently, the range is   
rdfs:Literal. This seems wrong - I expect that all clients expect it  
to be a string. It certainly can't be an xsd:float, which is licensed  
by this range.

It seems to me that OWL-DL could handle the special case of  
InverseFunctional string datatype properties by treating them as  
individuals named within a reserved namespace, and considering these  
properties to be in a class like "simple" properties, which  
disallowed any other constructs that could result in the range of the  
property becoming finite.

Essentially, DatatypeProperty(foo range(xsd:string)  
InverseFunctional) creates an additional, internal hidden property

ObjectProperty(*foo range(owl:stringObject) InverseFunctional) and

Individual(bar value(foo "12345")) generates an additional fact

Individual(bar value(foo <http://owl/stringobject/12345>))

Although this doesn't solve if datatypes in general, it would, I  
think, solve the problem for the 50M or so foaf files, and countless  
others that want *really simple* keys.

-Alan

> , but the solution is for OWL DL to figure out how to deal with  
> inverseFunctional datatypes, not to refit the 50M or so FOAF files  
> out there with "objects" that are really datatypes and should be  
> datatypes - this issue was the one that split OWL DL from OWL FULL  
> in the Owl 1.0 group, and remains unresolved.  I know there has  
> been work in the DL community on providing "key" functionality, and  
> Evren Sirin had a usable handling of it in his thesis work, but as  
> best I can read the OWL 1.1 specs, it doesn't seem to have made it in.
>  -JH
>
>
> On Nov 23, 2007, at 8:45 PM, Alan Ruttenberg wrote:
>
>>
>> For foaf, foaf-a-matic(http://www.ldodds.com/foaf/foaf-a-matic)   
>> writes
>>
>> <foaf:workplaceHomepage rdf:resource="http:// 
>> sciencecommons.org/"></foaf:workplaceHomepage>
>>
>> The species validator considers this OWL Full because
>>
>> Untyped Individual: Assuming http://sciencecommons.org/ is an  
>> individual
>>
>> Elsewhere, we have that the domain of foaf:workplaceHomepage is  
>> foaf:Document
>>
>> To get this to validate as DL(*)  I need to
>>
>> <foaf:workplaceHomepage><owl:Individual rdf:about="http:// 
>> sciencecommons.org/"/></foaf:workplaceHomepage>
>>
>> This is an annoying bit of incompatibility with RDF. Any way to  
>> fix it?
>>
>> -Alan
>>
>> (*) If the foaf ontology were itself valid as OWL DL.
>>
>> It's not that far off...
>>
>> Invalid Range Restriction: foaf:myersBriggs (ObjectProperty) has  
>> rdfs:range rdfs:Literal (Datatype)
>> (looks like a foaf bug - foafers: take note)
>>
>> Invalid SubProperty Axiom: rdfs:subPropertyOf is used with  
>> foaf:name (DatatypeProperty) and rdfs:label (AnnotationProperty)
>> (seems this could be handled on the OWL side - harm from the  
>> subproperty going the other way, but not from Object/Datatype  
>> property up to Annotation...)
>>
>> Multiple Types: Resource foaf:icqChatID is defined both as  
>> DatatypeProperty and as InverseFunctionalProperty
>> Multiple Types: Resource foaf:jabberID is defined both as  
>> DatatypeProperty and as InverseFunctionalProperty
>> Multiple Types: Resource foaf:yahooChatID is defined both as  
>> DatatypeProperty and as InverseFunctionalProperty
>> Multiple Types: Resource foaf:msnChatID is defined both as  
>> DatatypeProperty and as InverseFunctionalProperty
>> Multiple Types: Resource foaf:aimChatID is defined both as  
>> DatatypeProperty and as InverseFunctionalProperty
>> Multiple Types: Resource foaf:mbox_sha1sum is defined both as  
>> DatatypeProperty and as InverseFunctionalProperty
>>
>> All of these could be reasonably handled by making them objects. I  
>> won't hold my breath. For all but the sha1sum, it seems there is a  
>> reasonable analogy to mailto:
>>
>> -Alan
>>
>> ps. foafers: the link named foaf-dev@lists.foaf-project.org on  
>> http://xmlns.com/foaf/spec/ is to http://lists.foaf-project.org/ 
>> mailman/listinfo rather than to the expected mailto:foaf- 
>> dev@lists.foaf-project.org
>>
>>
>>
>
> "If we knew what we were doing, it wouldn't be called research,  
> would it?." - Albert Einstein
>
> Prof James Hendler				http://www.cs.rpi.edu/~hendler
> Tetherless World Constellation Chair
> Computer Science Dept
> Rensselaer Polytechnic Institute, Troy NY 12180
>
>
>
>