Re: UBL Naming Conventions & RDF

John--

To be more precise, the subject of this thread should be "UBL Naming 
Conventions & RDF *Naming Conventions* ".  The reason is that using 
predicates such as "hasParent" isn't hardwired into RDF, it's a naming 
convention that some people use (but others don't).  An obvious 
counterexample is that RDF defines "rdf:type", rather than 
"rdf:hasType".  Of course, the "hasFoo" approach has the merit of 
indicating the directionality of the relationship (at least it does to a 
human reading it).

Best practice in this area is, in my opinion anyway, still evolving. 
While you can certainly raise this issue on this list, it seems to me 
this would also be relevant on the Best Practices WG list as well 
(public-swbp-wg@w3.org).

--Frank

John McClure wrote:
> Hello,
> I am developing an ontology for legal documents that is based fairly 
> extensively on Word Net and also previous work that I did for the Data 
> Consortium. The ontology seeks to be directly compatible with the Dublin 
> Core, with the North American Industrial Classification System (NAICS) 
> and with the UBL Naming Conventions -- it is this last objective that 
> brings me to write this group.
> 
> The problem I am confronting is that the RDF normally defines 
> predicates such as "hasParent", a construct that conflics with UBL 
> Naming Conventions which state that "/[GNR7] UBL XML element, attribute 
> and type names MUST be in singular form unless the concept itself is 
> plural/".
> 
> Clearly, UBL naming rules require every XML element to be a noun, 
> qualified or not. In UBL there is no allowance for predicates like 
> "hasParent". It is therefore difficult to reconcile UBL naming 
> conventions with those in OWL-based ontologies. I think that if UBL 
> could be stated using RDF, then finally the two communities could 
> productively come together. 
> 
> So, my task has been to "take apart" the RDF's predicate construct, 
> distinguishing between predicate verbs and predicate nouns. A predicate 
> verb is simply <has> while a predicate noun is (or could be) <Parent>. 
> So rather than <Person><hasParent rdf:resource='uri'/></Person>, I am 
> suggesting that the pattern <Person><has><Parent 
> rdf:about='uri'/></has></Person> be used instead. Many advantages are 
> apparent from this approach, while there appears to be few or 
> no disadvantages.
> 
> Here's my question: is this the correct group with whom I should engage 
> on this question? Any references to relevant discussions in the past 
> about this would be appreciated -- I haven't found any in the archives 
> when I search for "Direct Object", "Predicate Noun", "UBL" or other 
> related keywords.
> 
> *Thanks,
> John McClure*
>

Received on Friday, 8 July 2005 15:10:08 UTC