- From: Jonathan Borden <jonathan@openhealth.org>
- Date: Fri, 11 Oct 2002 07:36:59 -0400
- To: "David Orchard" <dorchard@bea.com>, "'Roy T. Fielding'" <fielding@apache.org>
- Cc: <www-tag@w3.org>
David Orchard wrote:
>
> There are 2 assertions you made that I really don't get: 1) shouldn't
change
> software if resource becomes dereferencable; and 2) no implication that
> resources identified by dereferencable schemes will actually be
> dereferenced. I'll explain my confusion on these assertions in reverse
> order.
There's alot of discussion on various RDF lists regarding the notion of
"monotonic" vs. "non-monotonic" logic. RDF and subsequent specifications
e.g. OWL are being carefully crafted so as not to depend on any network
addressable representation being available for example:
<rdf:RDF>
...
<owl:imports rdf:resource="http:// ..../foo.owl" />
in this circumstance it is important that any conclusions(entailments) drawn
as a results of a successful import are monotonic, that is you can find out
_more_ about something, but nothing that you concluded was true becomes
false.
Software on the web needs to behave in a monotonic fashion and not depend on
URIs being dereferencable ... e.g. "please try again later" or "waiting for
transaction to succeed" rather than "your credit card is bad and hence you
are a criminal"
Jonathan
Received on Friday, 11 October 2002 07:55:34 UTC