head	1.1;
access;
symbols;
locks; strict;
comment	@# @;


1.1
date	2006.05.04.20.05.27;	author RicardoPereira;	state Exp;
branches;
next	;


desc
@
.
@


1.1
log
@Initial revision
@
text
@---++ Use PURLs as GUIDs

From the [[http://wiki.tdwg.org/twiki/bin/view/TAG/UsePURLsAsGUIDs][TDWG TAG Wiki]] by GregorHagedorn.

If using LSIDs causes problems with automated reasoning using semantic web technologies (see WhyWeShouldNotUseLSIDs), what arguments do we really have to not use PURLs? Roger Hyam mentioned at TAG-1 that he proposed PURLs at the GUID meeting, but that this was rejected. Why?

Do we need to use purl.org or can we set up our own purl resolver? I would not mind a purl.gbif.net, but we may need to reserve an appropriate DNS name, which would be independent enough to survive name changes both of TDWG or GBIF.

One advantage of LSIDs is their separation between data and metadata. However, it seems to me that this can also be achieved by content negotiation. With an accept: xml-tdgwmetadata we may configure the resolver to return metadata instead of the data.

The issue that purls in practice seem to be not very stable is truly a management problem, not a technical problem. Using different technology like LSIDs helps management to separate the issue of low management normal URLs and managed persistent purls, but I believe going through an agreed purl yields very similar results in documenting management decisions. The real problem with Purls as it appears to me, it that most institutions decide: well if it is simply URL forwarding and rewriting, then we simply do this ourself - but then do not follow this decision by allocating enough resources to truly hold the promise.

If we can agree on a URL-pattern that signifies the decision to keep a URL persistent, the GBIF could offer a service of notifying providers that they do not keep the contract. This seems to be a natural extension of the GBIF indexing service.

GregorHagedorn - 28 Apr 2006

Funny, today we were discussing about this approach in the TAPIR group. I would like just to add an example to Gregor's comments so that people not aware of what a PURL is can get an idea. I registered today an specimen:

http://purl.oclc.org/NET/BGBM/training/Specimen/1352

This URL is redirected to a BioCASe wrapper with a search by Specimen ID embed as a GET parameter. I do not paste the real URL because in the case of BioCASe it looks very ugly (just XML URL encoded). The wrapper is also returning the specimen described in ABCD 1.2 + the BioCASe protocol envelope, what is ugle, but the BioCASe was not created with this idea in mind (in TAPIR you can disable the protocol envelope).

JavierTorre - 28 Apr 2006
@