Re: ROR - Resources of a Resource

in response to Arjan Wekking <a.wekking@synantics.nl>

> Interesting schema, I've been touching the same kind of 'problem' 
> (describing parts of a website) but with the intent to use the 
> description as a real internal sitemap for a website. HTTP requests 
> would be looked up in this and the described resources would have other 
> properties that define how the page would be rendered to HTML etc.

I was going in that direction initially, but then decided to provide a way to distinguish the resource structure from the view into that structrure (which is what a sitemap really is). Simple sites will not necessarily need to provide a sitemap, but more sofisticated sites will want one, or several ones even (one for each view). 

> Anyway, I've taken a quick peek at your spec and one thing strikes me 
> as odd; why do you define the type of the described resources as a 
> literal? Why not make the those classes (like SiteMap, Menu, Contact, 
> etc) _real_ rdfs classes, and define the type using rdf:type?

> I see you mention the type literal being similar to rdf:type and seem 
> to support both, but why not just use rdf:type? Also, how does the spec 
> cope with rdf:lang in type literals?

The reason for this is simple: to lower the RDF learning curve. ROR is an XML and RDF/XML vocabulary. It is designed to be very easy to use by non-RDF people, while still providing all the benefits of RDF (and RDF vocabularies) to those who are comfortable with it. As people start experimenting with RDF through ROR, they will soon appreciate the power of RDF and want to add other vocabularies to their ROR documents. 

The type property will not support rdf:lang, since it does not represent a descriptive aspect of the resource. Only defined class names can be used in the type property (e.g. Main, Product, SiteMap, etc).

Hope this helps.

Dom Vonarburg
http://www.rorweb.com