[XML-DEV Mailing List Archive Home] [By Thread] [By Date] [Recent Entries] [Reply To This Message] Re: RE: A valuable lesson on the difference between XMLSchemas
Roger, I'm actually coming to believe that the nature of data modeling is even deeper than that. I've been thinking hard about the nature of resources, qnames and ontologies, and especially how they relate to REST.
One of the more intriguing results of this exploration is that if you do treat a qname as an identifier for a resource that can be addressed over the web, then among other things you have the ability to deduce a model of that resource through reflection and introspection. Additionally, it becomes possible to deal with the resource not as a static entity but as a dynamic one, with both volatile properties and associated resources. This to me has always been the Achilles Heel of traditional RDF - it assumes that data models themselves are static, when in fact entities undergo state changes all the time - one of the fundamental lessons of REST - not only of individual properties but also even of what specifically is modeled.
I'm currently working on an SPL content modeling system for Reed Elsevier, and have introduced a RESTful architecture which incorporates revisioning (non-destructive PUTs, in effect), which has, among other things, forced me to think about how revisions fit into REST. A given resource, such as a contact or company, has an associated URL and by extension qualified name (qname), yet the resource in turn provides an interface for its associated revisions, which are themselves resources. The revisions are static - they are a given representation of state, and once created, revisions can never be updated, only superseded or derived from.
Once consequence of this, though, is that the data instance, if not necessarily the data model, is transitory. This is consistent with the web, but much of SemWeb is really built on the data models and corresponding instances that are fixed..
This is apropos of nothing, other than the idea that even OWL modeling needs to incorporate state transients before it becomes sufficiently descriptive. On the flip-side, both XSD schemas and OWL ontologies serve only to create bindings between qnames, the principle difference being that OWL binding predicates are generally more comprehensive than XSD ones are (XSD enables an implicit container/contained relationship and an ordering relationship, and XSD 1.1 provides additional constraint bindings). I think its disingenuous to say that schemas do not impute semantics to data model structures - they do establish relationships, which is a form of semantic binding, and they also provide a mechanism to bind a secondary human semantics through annotations. Whether XSD models are complete is a different issue, but this gets back to the ultimate role of what a data model does. A data model, any data model, is simply a reflection of a given resource's properties and relationships relative to a given application context. If the context model changes, so does the data model requirements, and as a consequence so do the properties/relationships that get mapped. Again, I think this gets back to too traditional a view in the semantics community about the fungibility of the application context, and is a blind spot that limits the utility of either schemas or RDF/OWL as a tool.
Kurt Cagle Invited Expert, XForms Working Group, W3C Managing Editor, XMLToday.org On Sun, Nov 6, 2011 at 8:48 AM, Costello, Roger L. <costello@mitre.org> wrote: Hi Folks,
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] |
PURCHASE STYLUS STUDIO ONLINE TODAY!Purchasing Stylus Studio from our online shop is Easy, Secure and Value Priced! Download The World's Best XML IDE!Accelerate XML development with our award-winning XML IDE - Download a free trial today! Subscribe in XML format
|