On 21 Nov 2005, at 10:36, Stefano Mazzocchi wrote:
> Alf Eaton wrote:
>> On 15 Nov 2005, at 08:51, Stefano Mazzocchi wrote:
>>> Alf Eaton wrote:
>>>> Stefano Mazzocchi wrote:
>>>>> I'd much rather invest on automatically dereferenceable URIs
>>>>> now and avoid paying the discoverability price (not only the
>>>>> technical one but also the social one!) later on.
>>>> Stefano,
>>>> Just as example, which dereferenceable URI would you use for the
>>>> book "The History of Love", ISBN 0393060349, so that everybody
>>>> would be able to use that URI as an identifier for that book?
>>>
>>> http://purl.org/isbn/0393060349
>> I was thinking about this again today, while reading Tim Berners-
>> Lee's essay "What do HTTP URI's identify?" <http://www.w3.org/
>> DesignIssues/HTTP-URI.html>. Now, I don't know if he still stands
>> by this point of view, but his point as I understand it is that
>> HTTP URIs are designed to identify documents, rather than abstract
>> (or physical) concepts - one conclusion being "the idea that a URI
>> identifies the thing the document is about doesn't work because we
>> can only use a URI to identify one thing and we have and already
>> do use it to identify documents on the web.". This is also what I
>> like about info URIs, where, for example, uri:info:pmid/16286011
>> identifies the abstract concept of a particular scientific paper,
>> rather than http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?
>> cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=16286011 which
>> produces a document with information *about* the paper.
>> That particular example is complicated by DOIs, where http://
>> dx.doi.org/example/1111111 can produce a representation of the
>> paper which is in fact the same thing represented by uri:info:doi/
>> example/1111111, but this only works for examples where the object
>> being referenced is available in its entirety in a digital form.
>> It doesn't work for cars, for example (as in Tim BL's example), or
>> holiday resorts, or restaurants. In those cases, it seems more
>> appropriate to use an identifier which *isn't* HTTP.
>> In your example above, http://purl.org/isbn/0393060349 points to a
>> document *about* the book, but it doesn't point to the book itself.
>
> We are back to square one: nobody here is saying that
> identification and localization are the same thing. They are not
> and there is no debate there.
>
> The argument on the table is whether or not the use of an
> identifier *as-is* as a locator is useful or not.
> The advantage of using HTTP-URIs as URLs is that you use the
> existing web infrastructure (DNS+HTTP) for the dereferencing.
> The advantage of using INFO-URIs is that they are "just"
> identifiers and they can be created and used without people
> worrying about a direct mapping on the resources they locate.
Yes, that makes sense.
> Alf, you are right, my URI was defective in the way that it does
> not differentiate between the page and the book, so here is a new one:
>
> http://purl.org/isbn/0393060349#book
>
> and when you fetch that HTTP-URI as a URL, the #book part is not
> sent by the browser to the server, and you content-negotiate, say,
> RDF, you get a bunch of statements and one will contain the above
> HTTP-URI, so you know what to do next.
Right, that works for a book, because the whole book is availlable as
a digital object, assuming that when you request
http://purl.org/isbn/
0393060349#book you get the text of the whole book. But what if
you're talking about a car?
alf.
Received on Mon Nov 21 2005 - 17:17:08 EST