On Tue, 13 Aug 1996, Rzepa, Henry wrote:
> Those of you who have been at it for weeks, please excuse a chemist
> entering at a late stage.
Heh, I think its still a very early stage Henry - welcome aboard!
> I was rather hoping the Hypermailed archived
> might have some metadata elements to help me search for the
> information I seek! But having only the subject lines to help me,
> please forgive me if this information has already been addressed in
> a previous query (the arcane art of putting metadata into subject lines
> is one few of us have mastered).
That's a damn good idea Henry; something to add to my growing list of
interesting things to do. I've not looked into the Hypermail code so I
don't know how amenable it will be to being hacked over to do this.
> What I imagine this robot would need therefore is metadata of the following
> type
>
> <META NAME = "DC.RELATION" TYPE = "CHILD" CONTENT =
> "http://www.ch.ic.ac.uk/atp.pdb">
>
> My problem is, can I link the metadata element "DC.FORM"
> CONTENT="chemical/x-pdb" to the "DC.RELATION" TYPE = "CHILD"
> element? Given that the very use of TYPE modifiers is still under
> active discussion, am I hoping for far to much to be achieved under the
> constraints of HTML 2.0?
Personally I'd say that the IMT of the hyperlinked chemical resource
should really appear in that resource's metadata, rather than in the
DC metadata embedded in the original document . Then the robot would
follow the Relation element's URL in the original document to pick up the
metadata for chemical resource. Of course the chemical resource might
not be able to carry its own metadata but the HTTP server that returned
the resource would already tell the robot what the IMT of the resource
is, as well as other interesting stuff such as size etc (it has to
otherwise web browsers wouldn't know what the MIME type of the resource
was without resorting to naughty file extension guessing games).
If the robot-of-the-future used the proposed META method rather than GET
then it might even be able to pick up a Warwick Framework encapsulated
bundle of metadata related to the chemical resource, even though that
resource's file format does not permit embedded metadata.
However if you really, really want to have the IMT of the child resource
included in the Relation element of the parent document, we could just
dream up two new subelements called IMTmajor and IMTminor to hold the IMT
values. Then the HTML 2.0 embedded version of the above Relation (in my
prefered "dot.kludge" syntax) is:
<META NAME="DC.relation.TYPE.child.IMTmajor.chemical.IMTminor.x-pdb"
CONTENT="http://www.ch.ic.ac.uk/atp.pdb">
It would work but we're starting to get a subelement explosion after
constraining the element set size to thirteen... :-(
> Put simply, I want metadata in a chemically oriented HTML document
> to state that it contains a link to a document with precise chemically
> parsable syntax. I might mention that many of the chemical IMT formats
> in use could NOT be adapted to have metadata elements inserted directly into
> their content, and that some sort of LINK would be needed to make this
> association
Its not just chemical resources; some image formats and things like
spreadsheet files, etc can't hold much in the way of metadata. I think
that its important to remember that the embedded DC metadata in HTML is
just supposed to be (a subset of) the metadata for the HTML document
itself. Warwick Framework is intended for more general purpose metadata
transportation and can carry (or point at) arbitrary packages of
metadata.
Tatty bye,
Jim'll
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Jon "Jim'll" Knight, Researcher, Sysop and General Dogsbody, Dept. Computer
Studies, Loughborough University of Technology, Leics., ENGLAND. LE11 3TU.
* I've found I now dream in Perl. More worryingly, I enjoy those dreams. *
