On 5/18/11 7:08 PM, glenn mcdonald wrote:
it's not feasible, nor enforceable, nor desirable to develop
ontologies entirely with random URIs as identifiers.
Just to be clear, I said "pure identifiers", not "random URIs". I like
integers as local IDs. Add a base URI and you've got perfectly good
URIs for everything. Or a default prefix, if extra colons make you
feel more comfortable.
We're talking about practice, so "enforceable" is not the issue here,
and "desirable" is the question we're considering. It's certainly
feasible, though: countless data-systems throughout technological
history have used pure identifiers for machine purposes and
human-readable names for human purposes. It's not feasible without
tool support, I'll give you that. But neither are spreadsheets. I see
no reason to stipulate, in 2011, that the world-wide database should
be written in text editors.
Great point re. world-wide dbms and text editors as prime editing tool!!
This certainly exposes another aspect of the "elephant in the room"
problem.
Making it easy to scribble data on digital magic paper as a convenient
conduit to the global linked data space (that is the Web) is great, but
said quest is nothing more than a pattern amongst many. In a sense, the
one-size fits all issue is now finding its way into the Linked Data
meme, courtesy of fissure points in the overall narrative.
--
Regards,
Kingsley Idehen
President& CEO
OpenLink Software
Web: http://www.openlinksw.com
Weblog: http://www.openlinksw.com/blog/~kidehen
Twitter/Identi.ca: kidehen
