This would seem to me to be true if this were only well-formed XML but
isn't it the case that the schema defines the semantics where one
exists and not a particular instance of the application.
If one were deriving a schema from an instance, then clearly the
schema derived from the first example would be different from that
produced by the second.
But, as you imply in your first sentence, this is only a theoretical
conversation because we do have a schema that defines the semantics.
However, the real question on the table is- ought the schema be
defining rather than reflecting the semantics at all in an environment
where the they are
more properly defined elsewhere, viz. cataloging protocols?
Michael
On Mar 28, 2012, at 7:41 PM, Brian Sheppard
<[log in to unmask]> wrote:
> I think it's the reverse. Without any reference to accepted practice
> – or even to EAD markup in particular – the markup included
> implies in the first example, that there's some conceptual entity 'd
> id' that has as it's content something called 'unitdate.' The conc
> ept of unittitle doesn't exist.
>
> In the second example, 'did' contains 'unittitle,' the content of
> which is 'unitdate.' So the value of unittitle is the value of it's
> child: "1915-1920."
>
> This seems pretty obvious to my mind, and it illustrates the
> difference between semantic markup versus markup that suits some
> other purpose, like formatting or just delimiting data fields, say.
>
> I would agree, though, that these have no semantic difference:
>
> <did>
> <unitdate>1915-1920</unitdate>
> </did>
> ----------
> <did>
> <unittitle/><unitdate>1915-1920</unitdate>
> <did>
>
> Both share the same value and both have the same hierarchy.
>
> On Mar 28, 2012, at 8:42 AM, Michael Fox wrote:
>
>> There is no semantic difference between
>>
>> <did>
>> <unitdate>1915-1920</unitdate>
>> </did>
>>
>> and
>>
>> <did>
>> <unittitle><unitdate>1915-1920</unitdate></unittitle>
>> <did>
>>
>> except that your institution has unilaterally decided there is one
>> as you explained. That's fine but it does not signify in an
>> exchange enviroment where others do not know or share your
>> convention.
>
> --------------------------------------------------
> Brian Sheppard
> University of Wisconsin Digital Collections Center
> [log in to unmask] (608) 262-3349
|