After our dialogue of yesterday I again thought about the best term for the "pattern
name" in
http://www.orthuber.com/wp1.pdf which is simultaneously a identifier and a
address. Because of chapter 2.1 of
http://www4.wiwiss.fu-berlin.de/bizer/pub/LinkedDataTutorial/
the term "http URI" seems to be appropriate.
We have seen, that a complete, precise, and at once accessible definition in
exactly one place on the web can
be very helpful to avoid misunderstandings, and can save much time. Therefore the
identifier (the "pattern
name") should not only identify (e.g. the meaning of some numbers), is should
also point to (a file which
points to) all defining information (of this which should be identified, e.g.
of numbers).
If numeric web search (similarity search) should be integrated into the semantic web and/or if there is
interest in efficient representation of quantifiable objects, I would suggest to determine the concrete design
in a meeting, e.g. a workshop.
Wolfgang
----- Original Message -----
From: "Kingsley Idehen" <[email protected]>
To: "Wolfgang Orthuber" <[email protected]>
Cc: "Dan Brickley" <[email protected]>; "semantic-web" <[email protected]>;
"Linked Data community"
<[email protected]>
Sent: Tuesday, May 26, 2009 4:57 PM
Subject: Re: numeric web search (Was: URLs instead of URNs)
Wolfgang Orthuber wrote:
We know that a URL refers to a (unique) web address. If also
A URL is a Web Address based Identifier
then the Web Address determines also the URL. Because the Web address is
globally unique, the URL is unique
and can be used as unique identifier.
Is this correct?
The URL can be used as an Identifier because you can use a globally unique
Resource Location/Address as a
Name for a Thing (e.g. a Document), albeit with implications (i.e. mobility of
the Thing you name).
(then I could write that the pattern name in http://www.orthuber.com/wp1.pdf is
a URL, because it is based
on
the location of a unique "linking file" which points to all defining
information)
<http://www.orthuber.com/wp1.pdf> is the Web Address constrained URI (nee. URL)
for the resource: wp1.pdf
exposed to the Web via an HTTP server. I've made no mention of "all defining
information" .
Kingsley
Wolfgang
----- Original Message ----- From: "Kingsley Idehen" <[email protected]>
To: "Wolfgang Orthuber" <[email protected]>
Cc: "Dan Brickley" <[email protected]>; "semantic-web" <[email protected]>;
"Linked Data community"
<[email protected]>
Sent: Tuesday, May 26, 2009 3:04 PM
Subject: Re: numeric web search (Was: URLs instead of URNs)
Wolfgang Orthuber wrote:
Dan,
can a http URI refer transiently or accidentally to some address?
Of course.
Which term do you suggest for something which permanently refers to a (unique,
permanent) web address,
and
which differs if and only if the web address differs?
A URI that carries location/address specificity or dependency (transiently or
accidentally).
An Identifier with endowed location specificity (overtly or covertly) isn't
optimal, but that doesn't stop
it being an identifier.
A URL is a Web Address based Identifier -- a URI :-)
Kingsley
Wolfgang
----- Original Message ----- From: "Dan Brickley" <[email protected]>
To: "Wolfgang Orthuber" <[email protected]>
Cc: "semantic-web" <[email protected]>; "Linked Data community"
<[email protected]>
Sent: Tuesday, May 26, 2009 1:31 PM
Subject: Re: numeric web search (Was: URLs instead of URNs)
On 26/5/09 15:17, Wolfgang Orthuber wrote:
Dan,
in http://www.w3.org/TR/uri-clarification/ I read "An http URI is a URL"
. So I concluded that a different http URI is a different URL (address).
At this I assumed, that all http URIs which refer to the same address
(case insensitive), are defined as "identical". Is this correct?
I'd rather they'd have said "URL" is a technically obsolete but common
colloquial term for http and
http-like URIs. Identity of identifiers is tricky because you have to try to
distinguish between
identifiers which accidentally of transiently refer to the same thing, versus
those where it is built-in
to the definition of the scheme (eg. the port 80 and domain name
canonicalisation rules).
Dan
--
Regards,
Kingsley Idehen Weblog: http://www.openlinksw.com/blog/~kidehen
President & CEO OpenLink Software Web: http://www.openlinksw.com
--
Regards,
Kingsley Idehen Weblog: http://www.openlinksw.com/blog/~kidehen
President & CEO OpenLink Software Web: http://www.openlinksw.com
