How would a semantic web of bib data work?

So - I said I'd follow up my last post with my understanding of a
'semantic web of bib data'

I'm not sure I still understand RDF quite enough but to have a stab at
this, rather than cataloguing:

Harry Lindgen's Geometric dissections
Has an expression of
An original text titled 'Geometric dissections'
Has a manifestation of
The book published in 1964 by Van Nostrand
Which we have as item
bc987654321

(which is kind of what we do at the moment)

We would catalogue (for example)

http://somecatalogue.org/expression_id1234
Has a manifestation of
http://mycatalogue.org/manifestation_id9876
Which we have as an item
bc987654321

At somecatalogue.org, there would already exist catalogue record stating
that:

http://somecatalogue.org/work_id1234
Has an expression of
http://somecatalogue.org/expression_id1234

Of course, you would have to be happy that somecatalogue.org was
relatively authorative and had permanent URIs for it's
works/expressions/manifestations (in the example above I've assumed that
I am the first to catalogue this particular manifestation).

A semantic web search engine would crawl through such RDF statements,
and index all the related information. If we imagine I'm running a local
search engine, I'd start by pointing it at the root of my catalogue -
its homepage as it were. As it crawled through the data in my catalogue,
it would find references to somecatalogue.org, and would crawl to the
referenced URIs and index the data it found there. Let's assume my
crawler would allow me to set limitations on it's crawling - e.g. crawl
all references until you find the work record to index. If I want to be
the 'Google' of the bib world however, I would allow the crawler to
crawl all possible linked records - so having crawled 'up' to the work
record at somecatalogue.org, it would then crawl out from there, finding
all expressions linked from there, and thus out to all manifestations
that are linked back to this same work.

Practical problems aside, is this how we could expect a semantic web of
bib data to work, or am I going in the wrong direction?

Owen

Owen Stephens
Assistant Director: e-Strategy and Information Resources
Imperial College London Library
Imperial College London
South Kensington
London SW7 2AZ


Tel: 020 7594 8829
Email: o.stephens_at_imperial.ac.uk