W3C home > Mailing lists > Public > uri@w3.org > December 2007

Re: URIs & Namespaces

From: Erik Wilde <dret@berkeley.edu>
Date: Mon, 10 Dec 2007 00:24:38 -0800
Message-ID: <475CF7C6.5010403@berkeley.edu>
To: uri@w3.org
CC: Mike Schinkel <mikeschinkel@gmail.com>

hi mike.

just a short reply.

> Then your use-case for "place" is different than mine.  In my use-case, if
> it has a URL on Wikipedia, it can have a URL assigned.  Kinda by definition,
> no?  Those are the places I'm interested in.  If it is not important enough
> of have a Wikipedia entry, it's not important enough for me to catalog it
> (at the moment, anyway.)

sure, for a big vocabulary to make sense, it must rely on some way how 
to establish "importance". wikipedia could be one source for that. btw, 
we plan to load a big xml database with wikipedia in the spring and then 
start playing around to see how much semantics we can get out of there. 
if we are able to clean up the rather messy wiki-markup that is inside 
wikipedia's "xml"...

> I can appreciate yours; hopefully you can appreciate mine.  

sure. in a way, you are just a namespace for me. just like the getty 
thesaurus. or the u.s. zip codes. or smaller social vocabularies. i 
don't care about the size of place name vocabularies, i just want to be 
able to identify them, so that they can be used. so what would you 
prefer as a uri for your descriptions:

http://example.org/unitedstates/california/berkeley/southhall
location:southhall;http://berkeley.edu/campus
mikesscheme:southhall

where the latter would imply some way how you ensure that your names are 
  unique and can be identified by name only.

>> but i doubt that
>> a) any vocabulary can reasonably assume to be complete and 
>> the one everybody is using, and
>> b) there is one universal way of defining vocabularies that 
>> works well for everybody defining them.
> Those beliefs are shared by many people, and I agree in principle though not
> in all contexts. I think there are contexts in which it is applicable to
> find a single vocabulary.  If you believe that there is never a value for a
> single vocabularty then yours is definitely an ideology (even if you are
> somehow found to be right; ideologies do not have to be wrong.)  If you are
> open to the possibility that your stated beliefs are open to interpretation,
> then it is less of an ideology and might not be at all.  

of course it would be good if there were not too many namespaces, or at 
least some reasonable number of them, but i don't want to create a place 
name concept that somehow imposes limitations in that dimension.

so, the two "single vocabularies" i have foreseen so far (because they 
are the only ones for which i can see some real global consensus) are

- wgs84 coordinates
- country names

both of these vocabularies also have their problems, but i think they 
are at least something to start with. from there, it is all downhill in 
terms of how "shared" these vocabularies really would be. but as i said, 
i can fully understand your goal and wanted to look into "locations in 
wikipedia" myself (thus the xml database plans for spring). but i would 
not go as far as to call a wikipedia-derived place name vocabulary an 
excellent candidate for a "single vocabulary", but this is just a matter 
of taste and scale.

(as a side note: it would be interesting to check and see how many of 
the getty or alexandria place names are not in wikipedia. my guess would 
be that this would be quite a few, but i might be totally wrong.)

> But I am wordsmithing, and I probably shouldn't because it's tangential to
> the point of discussion.

in a way it's not, and i really like this level of discussion. but i 
think in a way our ideas are compatible, we are just thinking about 
slightly different levels of the problem of how to name things.

> It would be helpful, at least to me, if you gave some examples that you are
> considering to help me (and maybe others) visual what you are trying to
> accomplish.

the original (blog) post contains these examples, it lists the four 
different solutions that i could come up with (the fifth one would be 
the conclusion that something like that should not be done).

http://dret.typepad.com/dretblog/2007/12/uris-namespaces.html

cheers,

dret.
Received on Monday, 10 December 2007 08:24:56 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:25:11 UTC