[rdfweb-dev] minor spec update (saturday)

Morten Frederiksen mof-rdf at mfd-consult.dk
Wed Mar 17 17:50:02 UTC 2004


On Wednesday 17 March 2004 14:35, Dan Brickley wrote:
> Hi. Just to let folks know... I fixed up the spec management
> scripts on saturday and made a minor update to the spec, adding
> foaf:primaryTopic, foaf:PersonalProfileDocument and foaf:tipjar
> terms to the vocabulary. I hadn't announced it yet as there are a few
> more bugs to fix, but since word has gotten out! ...
A few comments:

foaf:PersonalProfileDocument:
The prose says "There is just one foaf:Person described in the document", but 
I assume the meaning is that there's is only one *authoritative* description 
of a Person, as otherwise it'd be impossible to indicate foaf:knows 
relationships etc.?
Also: "Anything that is a foaf:Person and that is the foaf:maker of some 
foaf:Document will be the foaf:primaryTopic of that foaf:Document." This 
seems plain wrong, as then it'd be impossible to use foaf:maker for anything 
but PPDs.
The last paragraph seems to contradict the purpose of the class: "Note that a 
foaf:PersonalProfileDocument will have some representation as RDF. Typically 
this will be in W3C's RDF/XML syntax, however we leave open the possibility 
for the use of other notations, or representational conventions including 
automated transformations from HTML (GRDDL spec for one such technique)."
If this class of documents is extended to possibly cover non-RDF(/XML) 
documents, we no longer will have any advantage over foaf:Document, that is, 
one can no longer expect a PPD to be RDF(/XML).
In addition to this, the prose seems to leave it "optional" whether to use 
foaf:maker/foaf:primaryTopic ("this can be inferred"), but I don't see how a 
PPD without at least one of these properties solves the problem of 
identifying the person that is described.

foaf:primaryTopic:
The prose says it's a owl:FunctionalProperty, and the schema as well, but 
it's not marked as such in the "header" of the term definition.

foaf:tipjar:
Leigh outlined some problems with the chosen approach, perhaps an example 
outlining how to extend the description of a foaf:tipjar object with a 
description of an imaginary payment method would help towards accepting that 
this is a step towards auto-payment, not a blind alley?


Aside: Any chance that all the ranges that are currently marked as 
rdfs:Resource could be changed to owl:Thing (where applicable, like 
foaf:depicts and foaf:made)? As it is, rdfs:Literal is a subPropertyOf 
rdfs:Resource, so there's really no value in this. It would also help with 
"validation".


Regards,
Morten



More information about the foaf-dev mailing list