Chuck,

I believe that the resolution protocol needs to be established or the identifier scheme will not persist. In the SGML days, many people promulgated the use of Formal Public Identifiers (FPI), which can still be seen in use in the DOCTYPE declaration of HTML documents and the like. FPIs defined a syntax for the identifier, but did not specify a resolution mechanism. Rather it left resolution loose so that the identifiers would not become stale. Eventually a catalog system for resolution evolved that was in fairly common use (and was popularized with XCatalog), but I think FPIs ultimately have not worked out precisely because the resolution mechanism was not mandated in the specification.

One of the reasons I like LSIDs is that the specification both leaves the resolution mechanism open, saying that several protocols might be used and it might evolve, while at the same time fixing the current resolution protocol to use DNS as its base for finding a resolver. Using SRV records still permits the indirection that is needed for the indetifiers to persist indefinitely. Providing a concrete resolution protocol based on open internet standards allows an interoperable system to quickly materialize. And allowing for future additions and changes to the resolution protocol lends some degree of future-proofing the identifiers. All in all I think it is a reasonable balance where there are several conflicting requirements.

Matt

Chuck Miller wrote:

Do we want GUIDs that are dependent upon a specific protocol or schema?

Seems like the ideal would be an approach that disconnects them so that the decisions for GUIDs and for protocols and schemas can be made separately.

From what I've been hearing so far the GUID is embedded inside of something else, like LSID or DOI. That is, the unique identifier part is wrapped inside something that is understandable only by a specific protocol or locating mechanism. Is this unavoidable?

Chuck

Chuck Miller Chief Information Officer Missouri Botanical Garden 4344 Shaw Blvd St. Louis, MO 63119

-----Original Message----- From: Roger Hyam [mailto:roger@TDWG.ORG] Sent: Friday, November 18, 2005 4:40 AM To: TDWG-GUID@LISTSERV.NHM.KU.EDU Subject: [TDWG-GUID] RDF/architecture/ontology - migration

This is carrying on from Steve's comments under the "Taxon debate synthesis?" thread. I started a new thread as it seemed to be getting a little deep and no longer fitting the title.

Steve's comments are hitting the nail right on the head. From the conversations and thoughts I am having about architecture of TDWG standards all roads are leading to RDF - which is annoying because it makes writing balanced documents that compare the alternatives difficult :)

Steve outlined one of the most promising paths forward for TDWG standards. Taking this route is not a matter of just saying "lets do it all in RDF" there would be a long way to go if we went this way - but at least we would be doing things the same way the rest of the semantic web world and that means there are tools and people out there to help.

The one point Steve didn't stress is that RDF is the "bees knees" for data handshaking - i.e. combining data from different domains. This means that the extensibility and version problems that are our main hurdles as the moment will tend to go away. But have no fear there will be other problems to replace them.

I am comfortable talking about GUIDs in terms of moving towards representing TDWG data in RDF. It certainly makes more sense of the GUID discussions to me. But...

What we need to bear in mind is that there is a great deal of knowledge captured in XML Schema within the TDWG community and that knowledge (or at least the good bits of it) need to be migrated forwards. People have also invested a great deal of effort in developing XML Schemas and may be reluctant to move on.

This is all leading to TAG stuff rather than GUID stuff but as Steve and Rod and others point out the two are very closely connected - along with the protocol stuff...

How much should the GUID debate assume that we are using current XML Schema based standards and how much should it assume a move to an RDF style approach - or doesn't it matter?

Roger

--

---

Roger Hyam Technical Architect Taxonomic Databases Working Group

---

http://www.tdwg.org roger@tdwg.org

+44 1578 722782

-- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Matt Jones jones@nceas.ucsb.edu National Center for Ecological Analysis and Synthesis (NCEAS) UC Santa Barbara http://www.nceas.ucsb.edu/ecoinformatics ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~