Consistent Identification?

[Patrick Durusau <patrick@durusau.net>]

Greetings!

The current language on Consistent Identification reads:

> IRIs provide a well-developed infrastructure for globally unique 
> identification which also facilitates integration with network-based 
> solutions, and so should be used to identify resource descriptions. 
> These identifiers can then be used to link document content with 
> descriptions, or descriptions with other descriptions.
>
> This requirement facilitates the following goals: facilitate 
> innovation 
> <http://wiki.oasis-open.org/office/OpenDocumet_Metadata_Use_Cases_and_Requirements#goal-innovation> 
> and ensure interoperability 
> <http://wiki.oasis-open.org/office/OpenDocumet_Metadata_Use_Cases_and_Requirements#goal-interop>. 
>
>
No, IRIs facilitate integration with RDF based solutions and not 
"network-based solutions."

Imagine for a moment that we have one of Gary's real estate transaction 
systems, none of which use IRIs for any purpose whatsoever. Within the 
interchange of those documents, the parties have identified metadata 
that is meaningful to them and have agreed protocols for its processing. 
There seems to me to be no reason to require that they use IRIs for 
"globally unique identification." Their current systems of 
identification suit their present purposes and forcing them to adopt a 
new system of identification simply imposes a barrier to adoption of ODF 
for their systems.

Nor does such a requirement "ensure interoperability." Recall that we 
have specifically discussed that ODF applications must preserve metadata 
that they do not understand. Can we really say that an ODF application 
that does not understand some portion of metadata is really 
interoperable with an ODF application that does? True enough, it doesn't 
step on the metadata that the other application understands but on the 
other hand, it doesn't use it.

Certainly we should say, for interoperability reasons, how metadata is 
associated with portions of an ODF document (whether that is structure 
or content). That means that there is a level playing floor for all ODF 
applications that if they do understand the metadata, there is a clear 
and consistent way for it to be applied.

Given that the variety of metadata that may be applied to ODF documents 
is unknown and probably unknowable, we are not competent to specify at 
this point the semantics and processing of that metadata. I would 
include in the processing of metadata the identification that here is 
suggested should be done by IRIs. If we are not competent to specify 
enough processing to make all metadata meaningful as defined by the ODF 
specification, I see no need to specify how such metadata should be 
addressed.

*NOTE* that is not to say that for metadata we do define we should not 
be able to specify that it must use IRIs, for example, because we are 
defining the metadata and to me that includes identifying how it will be 
identified. But that is solely for metadata that we define, not metadata 
that is defined by others.

We don't lose anything by allowing users in particular communities to 
identify their own metadata and avoid forcing people to adopt new 
identification systems for their metadata.

I have heard that Ingenta now has 200 million triples for its 
publications. Interesting but pales beside the number of triples that 
would be required to identify all the subjects within the documents in 
that single document collection. And recall that we are only talking 
about one document collection and not the various collections of other 
vendors who all have identifications systems of their own. Check out 
Swoogle if you want to see how "consistent" identification is using 
IRIs. If we all use different IRIs for identification of some subject we 
are simply back where we started except now we have one more 
identification system, IRIs, that has to be integrated with all the 
existing ones and new ones that are being invented all the time.

Yes, people argue that there will be convergence on IRIs for identifiers 
but it hasn't happened with a small community of people actively using 
IRIs as identifiers. In view of the current experience I would say there 
is no evidence for convergence at this point. Convergence strikes me as 
more an article of faith than fact and as such I would prefer to enable 
users of ODF to use any identifiers, including IRIs. Who knows? 
Convergence may happen and if so, ODF should be ready to use them. On 
the other hand, if it doesn't, ODF should be ready for that possibility 
as well.

Hope everyone is having a great day!

Patrick

-- 
Patrick Durusau
Patrick@Durusau.net
Chair, V1 - Text Processing: Office and Publishing Systems Interface
Co-Editor, ISO 13250, Topic Maps -- Reference Model
Member, Text Encoding Initiative Board of Directors, 2003-2005

Topic Maps: Human, not artificial, intelligence at work!