on 9/13/01 2:23 AM, Clayphan, Robina at [log in to unmask] wrote:
> My thoughts are: firstly you would repeat an element if you wanted to enter
> values of different types i.e. qualify the element with a scheme. For
> example - DC.Subject expressed as LCSH and a repeat for DDC. If the value
> of DC.Subject were simply going to be several keywords then repeats would be
> superfluous and all the words could go in one instance of the element.
>
> A second reason for using repeats could be where the values of an element
> need to be clearly differentiated because they refer to entities that are
> different in their own right - e.g. several Creators.
Robina didn't say if she is using RDF or not. I'll assume not and make this
a plug for the use of RDF's bag, sequence, and alternative containers.
The use of rdf:Alt would make it absolutely clear that the subjects
expressed as LCSH and DDC were indeed logical duplicates, the same
information encoded differently. Another example might be keywords in
different languages, rdf:Alt with xml:lang attributes makes this clear.
The use of an rdf:Bag for the simple case of unadorned keywords is also less
ambiguous than repeated simple elements and more portable than something
like a single semicolon separated list. It is also easier for general
purpose software to handle since you don't have to know what separator
happens to be in use.
In the case of dc:creator, use of rdf:Bag or rdf:Seq make it clear whether
there is any hierarchy among the authors. Use of rdf:Alt for a single
author within the bag or sequence can for example show Romanji and Kanji
forms for a Japanese author of a book published in the U.S. and Japan.
Note that formally in all of these examples there would be a single DC
property element. The value of the DC element would be the RDF container,
which holds multiple items.
Alan Lillich
Adobe Systems Incorporated.
|