I think you could (or at least I could) think of a schema as a model.
Consider that you can draw diagrams - boxes and lines- to represent n
idealized document. For simple document types, a diagram can even be
easier to understand than a DTD or XML Schema.
There is no real difference between such a diagram and a similar one for
a relational database or for an object model (or for an organization,
for that matter), except for some language-specific details. And I
would consider the relational or object diagram to be a "model", so why
not the one for the document grammar?
TomP
On 12/20/2016 12:25 PM, Henry S. Thompson wrote:
> John Cowan<> writes:
>
>> >It's mostly a grammar, but not exclusively. In particular, subtyping by
>> >extension and restriction are data-model-like features that have nothing to
>> >do with whether a particular document conforms to a particular grammar, as
>> >it is always possible to compile them out of the schema.
> I hesitate to disagree, but using OO-inspired techniques as a concise
> way to express multiple grammar rules doesn't mean the result isn't
> still a grammar -- it's just like the use of metarules in GPSG, a
> notational device. The explicit grammar is expanded by the
> interpretation/application of the metarules, producing an expanded
> (implicit) grammar, which is in turn what is used to determine
> language membership/schema validity.
