> Yes, that is a very distinct notion, even though I think that
> the digital "chunking" going on currently is making the
> distinction less and less relevant. For example, why is
> information about a picture metadata, but information about a
> paragraph, or a sentence, or a word, not metadata? All these
> can be included in a document. Some formats (HTML,
> typically) just stores the picture separately. I think that
> as soon as we want to treat an object as a separate entity
> from its container, an annotation is just metadata.
>
> However, there is an important distinction to be made between
> an object-as-is and an object-in-context (such as a word at a
> given point in a text)
Both are "metadata" but I do think the distinction between "cataloguing"
and "annotating" is important regardless of whether the annotiation is
within the data stream (ala TEI) or external (e.g for images or even
endnotes/footnotes in text).
> I hope my previous mail did not appear to suggest that. In my
> definition (and Andy's, I think?) structured data in itself
> is not metadata, until it is *about* something else.
This was an observation I made in 2000, so you are safe ;-).
However, the problem is that it is not always clear. Data by its very
nature is *about* something (even if it is about itself, which would
presumedly be self-referential metadata). Census data for example is
*about* population. As such, I'm not sure that you can have structured
data not about something (but please quote a counter example).
> I think metadata *is* a useful term. The terms cataloguing
> and annotating are also useful, but less for describing the
> information and more for describing the process or activity
> of producing the information.
My issue with "metadata" is that it is an overused term, and used in
different areas to mean sometimes very different things. So it becomes a
bit like "repositories" - you may end up with people attempting with two
people both doing "metadata" trying to work together and failing because
in fact they are doing two very different types of thing.
> What I mean is that machine-processability refers less to
> machine-processability of the contained data (which is also
> interesting of course), and more to machine-processability of
> the relation between the data and the resource. Even textual
> descriptions are metadata, as long as the relation to the
> described object is machine-processable.
OK, so metadata implies (or requires) an explicit (machine processable)
link between the data/object and the related metadata/object?
Matthew
|