I endorse Diane's support for separate elements.
My experience with importing heterogeneous metadata for use in American
Memory is that I would ALWAYS prefer separate elements from the point of
view of building an application to index and display the metadata.
I do realize that this can create complexities on the data entry side that
can't be robustly dealt with in a flat-file database. Now that XML-aware
tools are more available, I believe that going for the simplicity of a
flat-file database may prove a false economy in many cases.
Caroline Arms [log in to unmask]
Technical Coordinator for LC/Ameritech competition
National Digital Library Program
Library of Congress
On Thu, 13 Sep 2001, Diane I. Hillmann wrote:
> Robina:
>
> The reason I generally prefer separate elements (in addition to the ones
> you cite) is a very practical one--it's very difficult to enforce standard
> separating punctuation. If you have data where punctuation is used
> ambiguously, crosswalking, display or any other use of the data becomes
> quite problematic. Consider, for instance, what happens if LCSH headings
> using commas to enforce the order of words get mixed with commas separating
> actual headings. You'd have the same difficulty if your name data included
> corporate names with hierarchies separated by commas and then used the same
> separator for multiple instances. Lots of other examples, of course.
>
> A lot depends, of course, on how you're acquiring your data and how
> reliable the source, but in general, repeats are safer, in my view.
>
> Diane
>
|