I completely agree with Peter Graham's point that "parsimony can
add to complexity". I'd like to propose an acid test for the
Dublin Core:
We should be able to print out a page with the thirteen (or
however many) elements and give it to naive users to fill in
with no further instructions.
If they are mostly able to describe statistically average
materials (such as books, articles, and the like) by filling in
the blanks more or less correctly -- and if they usually do not
need to fill in any one blank more than once (for example, with
Author, Illustrator, and Publisher) -- then the Core is good.
The Dlib article on the Warwick workshop mentions statistics on
the use of IAFA template fields; perhaps the "top five" of these
should definitely be in the Dublin Core as separate fields. Does
anyone out there have these statistics?
Tom
> From: Peter Graham, Rutgers University Libraries
>
> In this context, by the way, I think parsimony can add to complexity, not
> reduce it. I'm not sure where the drive to reduce the number of elements
> comes from; thirteen certainly isn't too many, and I can see the number
> growing slightly rather than shrinking. As with the proposed elimination of
> "publisher" in favor of a variety of OtherAgent, the perceived economy is
> only a shifting of location in a way that in fact adds complexity (it's all
> very slight, of course, but that's what we're talking about). --pg
|