Tom Heath wrote:
Ed,
Thanks for asking the question - it's a good one, and I'd also be keen
to hear the background to this decision from Aaron. As you flag up in
your response re the use of #this, the issue here is much more about
distinguishing Districts from "Documents about Districts" than it is
about slashes or hashes.
Tom.
Tom,
Yes.
I see Peter is here :-) I've been trying to drive home this issue of
Identity and Address separation. The URI <> URL unless he "Thing" in
question is an information resource, and only then is appropriate to
take the URI/URL approach.
A Web of Data is a Web of Data Objects. The Document Web is a Web of
Documents. In either case you must uniquely identify the Objects :-)
#this is just a cheap mechanism for making a URI from a URL without the
cost of a 303 .
We have to do a better job of explaining what Object Identity is about
in the context of the Linekd Data Web, especially as this is an area of
computer science that predates the Web.
Kingsley
-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Ed Summers
Sent: 16 April 2008 00:33
To: Peter Ansell
Cc: public-lod@w3.org
Subject: Re: [Linking-open-data] watchdog.net and LOD best practices
On Tue, Apr 15, 2008 at 6:21 PM, Peter Ansell
<[EMAIL PROTECTED]> wrote:
The insistence on slash URI's having 303 is mostly
philosophical if
you are still going to allow hash URI's which people have no real
reason to go away from. When you consider that user agents
do not send
the hash part anyway, hence you have no idea at the server whether
they are wanting the resource or just part, so you send back the
result of a slash URL resolution... No need to push someone away
because they didn't follow and arbitrary rule for "non-information
resources".
I'm really not trying to push anyone away -- and I apologize
if it appeared that I was. My exposure to linking data has
largely been informed by watching discussions on here and
reading documents like How to Publish Linked Data on the Web
[1] and Cool URIs for the Semantic Web [2].
The value of these documents IMHO, is that the provide some
pretty clear guidance on URI design w/ httpRange-14 in mind.
I know Aaron is not a newcomer to the web or RDF -- so I was
genuinely curious about the thought process behind the service.
If there is any experimental evidence the Semantic Web
will actually
succeed if there are double the number of requests needed for a
single resolution then it may be interesting to revisit
the idea of
assuming people are bad semantic web citizens because they
implement
things pragmatically.
That's why the hash option is there. I'm fairly certain that
the watchdog rdf descriptions would just have to use a hash
URI in the description to be compliant--no double requests or
303 is necessary.
For example:
@prefix : <http://watchdog.net/about/api#> .
<http://watchdog.net/us/MD-08#this> a :District;
:almanac
<http://nationaljournal.com/pubs/almanac/2008/people/md/rep_md08.htm>;
:area_sqmi 307;
:center_lat 39.0231;
:center_lng -77.1421;
:cook_index "D+20";
:est_population 700364;
:est_population_year 2005;
:median_income 68306;
:name "MD-08";
:poverty_pct 6.2; :state <http://watchdog.net/us/md>;
:voting true; :wikipedia
<http://en.wikipedia.org/wiki/Maryland's_8th_congressional_district>;
:zoom_level 10;.
It would however be nice to have link rel="alternate"
type="application/rdf+xml" href="blah.xml" etc. in the
head of the
html though to facilitate the discovery of the RDF/N3/JSON
materials
automatically. That doesn't imply that the blah.xml has to 303
redirect to a third resource "just in case" though.
Right it doesn't have to. But the URI used to identify a
district would just need a hash part appended to it as above.
I'm not really trying to pedantic here, I'm trying to
understand the best practices myself.
//Ed
[1] http://www4.wiwiss.fu-berlin.de/bizer/pub/LinkedDataTutorial/
[2] http://www.w3.org/TR/cooluris/
--
Regards,
Kingsley Idehen Weblog: http://www.openlinksw.com/blog/~kidehen
President & CEO
OpenLink Software Web: http://www.openlinksw.com