On 5/19/11 10:32 AM, Whitley, Zachary C. wrote:
Can't we just say that there is a spectrum of methods for producing identifiers
that have certain tradeoffs and to use the method that best fits your use case?
Yes!
And that we have Identifiers that have varying behaviors driven by
pragmatic goals.
If you're in the machine readable camp you can just ignore any human
readability aspect of the identifier and just treat it as the monkey who just
randomly typed Shakespeare. Since there's a no unique names assumption we can
have multiple identifiers so why not both machine and human identifiers if
that's what you want with appropriate sameAs, equivalentClass and
equivalentProperty statements?
This is the conundrum, we are not talking about how you can have both
i.e., machine oriented identifiers for naming entities and human
(hack-able) identifiers for accessing their representations.
We are inadvertently doing a one size fits all by pushing for an
idealistic pattern when the target realm is duplicitous (Name/Address
ambiguity courtesy of HTTP scheme URIs for Entity Names and Data Object
Addresses).
--
Regards,
Kingsley Idehen
President& CEO
OpenLink Software
Web: http://www.openlinksw.com
Weblog: http://www.openlinksw.com/blog/~kidehen
Twitter/Identi.ca: kidehen
