On Fri, 24 Apr 1998, Rachel Heery wrote:
> On Fri, 24 Apr 1998, Dan Brickley wrote:
>
> > daunting complexities found in other languages like KIF or CycL. While RDF
> > is certainly not the last word in knowledge representation, it should
> > hopefully be enough to meet many of the requirements of the digital
> > libraries / metadata community.
>
> I think it would be helpful to outline some of the benefits that RDF
> offers for 'users' whether service providers or searchers. And to
> acknowledge that RDF is not proven technology, it is at the research
> stage. Note the recommendation at the start of the W3C specs that 'only
> experimental software be implemented to this document', it is very much
> work in progress.
Yes, this is exactly why I was soliciting feedback/input from eLib for the
schema document. If we wait until the spec reaches 'Recommendation' stage,
we lose the opportunity to contribute to its evolution.
> Some quick thoughts on benefits for 'resource operators' (service
> providers ... what shall we call them??):
Depends who you're talking about. Both primary content providers (eg.
digital image libraries, online journals) and internet cataloguing
services (eg. the eLib ANR projects) stand to gain from having a standard
framework for machine-readable resource description.
> The most obvious benefit RDF promises is a 'standard' syntax for
> describing web resources which can be co-located with the resource.
RDF currently provides two syntaxes; the interesting thing about RDF is
not the syntax but the fact that there is a single abstract model that can
be expressed in a variety of syntaxes. It provides a unified framework for
thinking about all varieties of metadata without the syntax dominating the
debate.
> This will make creating and gathering simple metadata created at source
> easier. This means it will be easier to create viable products for
> metadata creation.
Better still, it means that other people's generic RDF tools for metadata
creation may work pretty much 'off the shelf', so the need to create
metadata management tools on a per-project basis should be greatly
diminished.
> RDF also allows metadata to be located remotely from that resource. This
> enables creation of new sorts of metadata (for collections, images
> etc). It also offers the opportunity for third parties to build up
> collections of metadata referenced from the resources that sit on remote
> web sites.
Yes; Netscape 5's bookmark system is particularly interesting in this
respect. Your personal bookmark system in effects become an RDF database,
and can include entire sets of search results, site-maps and other RDF
resources acquired on the Web.
> Further down the line RDF offers more. It will allow reference from
> metadata to registries, which might or might not be stored as RDF schemas.
The notion of a centralised metadata 'registry' goes somewhat against the
architectural spirit of the web. The Web itself can be considered the
registry. Anyone with a reliable webserver can publish an RDF schema
document to describe the metadata vocabulary they've adopted. By making
the schema document machine-readable, users can precisely describe their
metadata without having to formally register anything with a centralised
registry.
> We might see registries of metadata sets (schemas)
> which provide authoritative versions of metadata formats,
> definitions of elements, allowed schemes, mappings between formats. This
> might aid validation at the metadata creation stage and the ability to
> automatically update collections of metadata.
Yes. None of these require centralised schema repositories though. Digital
signatures on RDF should be enough to enable the above features.
> (For those interested in this area I recommend Andrew Layman's paper to
> WWW7 'Specifying metadata standards for metadata tool configuration'
> available at following URL til the end of May
> URL: http://www7.conf.au/programme/fullpapers/1913/com1913.htm
This is a good read (although by Andrew _Waugh_ not Layman; Andrew Layman
is co-editor of RDF Schemas). Andrew lists a number of requirements for a
schema specification language. I'm hoping that the eLib community might be
able to come up with something similar...
> Well that is my understanding of RDF :-)
:-)
Dan
--
[log in to unmask]
Research and Development Unit tel: +44(0)117 9288478
Institute for Learning and Research Technology http://www.ilrt.bris.ac.uk/
University of Bristol, Bristol BS8 1TN, UK. fax: +44(0)117 9288473
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|