On Mon, 14 Apr 1997, Paul Miller wrote:
> On Mon, 14 Apr 1997, Sam Saunders wrote:
> >
> > The OCLC automatic indexing system is certainly wayward. Sending in
> > a couple of test documents yielded such odd descriptors that I
> > blush to think of the results. The meanings and purposes of a text
> > can obviously not be guessed at by a machine. It's difficult enough
> > even for a specialist, so I guess we're not out of business for a
> > year or two in offering professional indexing.
>
> Yes... it produced 'interesting' results for a number of AHDS projects,
> too... (did you know that Richard III was involved with ADS...?).
>
> Still, it's an interesting attempt at further automating metadata
> creation, and marks the next stage on from tools like mine
> (www.ncl.ac.uk/~napm1/dublin_core/ but about to move to the ADS) and the
> Nordic Metadata Project. It'll be interesting to see how this ocassionally
> quirky tool evolves following this process of public testing...
For info, there's a new, somewhat experimental, Dublin Core generator
available at
http://www.ukoln.ac.uk/cgi-bin/dcdot.pl
This is a hybrid of the two kinds of system mentioned above. Given a URL,
it retrieves the resource and generates some Dublin Core metadata
automatically - part of this process involves extracting existing metadata
from any META tags in the page. These values are then put into a Web
form, allowing subsequent modifications and additions to be made. Help,
in the form of a 'floating', context sensitive, help window is available
while making such modifications. (Please note that the help system uses
JavaScript and may therefore not work on all browsers).
As with the other tools, what you end up with is a set of HTML META tags
suitable for embedding in your Web page. These tags conform to the
recommended syntax for embedded Dublin Core agreed at the recent DC4
Workshop in Australia.
Those who are worried about such things might like to note that copies of
any metadata created using this tool will be stored here at UKOLN. At
this stage, this is primarily to enable monitoring of how well the system
is working. However, in the longer term it may be possible to use these
stored copies as the basis for metadata updates, either as people re-visit
the Web form to update their own metadata, or perhaps automatically by
looking for resources that have been updated since the metadata was
created.
Andy.
--
UK Office for Library and Information Networking
University of Bath, Bath, BA2 7AY, UK Voice: +44 1225 323933
http://www.ukoln.ac.uk/~lisap/ Fax: +44 1225 826838
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|