On 24 May 2005, at 21:42, George Macgregor wrote:
>
> I agree, Steve. Obviously, metadata does tend to assume familiar
> manifestations (i.e. author, title, publication date, subject, etc,
> etc.)
> because these attributes tend to be the most useful access points for
> most
> users and describe most information entities adequately (e.g.
> monographs,
> documents, reports, serials, websites, etc.).
In IT more generally, the most common manifestations of metadata are
things like "date modified", "modified by", and "size in bytes".
Metadata tends to be there primarily to facilitate processing and
auditing rather than discovery - most metadata in non-repository IT
systems tends to be invisible as far as users are concerned, and is
there mainly to improve the performance of the program. Thats a
generalization, of course.
(Information entities in this case being things like customers,
vehicles, purchase orders, accounts, sales records, stock levels,
policies, schedules, communications and all of the other things that
aren't books, articles, or learning objects :-)
Operating-system metadata is also an interesting category of metadata,
that is moving (well, on OS X at least) from traditional IT-style
performance metadata towards something more like resource description.
Library/info science tends to extend the definition of data to include
the properties of real-world objects, whereas IT people tend to confine
data to things in databases and files. Hence, from an IT-centric
definition, the title of a book in a catalog database is data, not
metadata, as it doesn't describe an object in the database, it is the
data itself.
Personally I'm not really that bothered either way about the
definitions ... but I am interested in what the Metadata SIG considers
to be its scope - is it just catalog-style records for learning
resources, or something a bit broader?
-S
|