Human-Opaque URIs

Hi All,

We've been using meaningfully-named URIs at work for a while, and
after 9 months I'm finding them difficult to keep totally persistent.

E.g. we store an inventory of applications, but every so often the
name of an application changes, and when it does the old URI doesn't
make sense any more. People (or rather automated RDF generation
systems) start using new URIs to represent the same thing, and it all
starts to decay a little. OWL should be able to take care of this, but
currently none of our systems have OWL inference engines.

The main problem isn't with the URIs themselves, but with the fact
that we've got into the habit of expecting to be able to read them
ourselves. When the URI text doesn't correspond to the 'thing'
anymore, it niggles.

I'm now considering using opaque numbers in URIs to represent things -
e.g. http://sw.example.com/2003/01/application/23 - and am wondering if
other people do this and what their experiences are.
In particular, what would be the advantages/disadvantages of working
in a world where URIs contain little human-readable information?

Cheers,

Phil

Received on Friday, 2 April 2004 04:59:29 UTC