[tdwg-tapir] Darwin & RDF

[Renato De Giovanni]

Thanks for all feedback.

Since there seems to be no big difference between using hash or slash, I
took the liberty to change the Darwin namespaces by adding a slash in the
end.

Although we could have different concept identifiers for RDF and TAPIR,
this could probably bring confusion in the future so I suggest we keep
them the same.

By using a slash, all concept identifiers for TAPIR remained the same and
we also followed one of the TAPIR recommendations: avoid using URL special
characters in concept identifiers. Otherwise we would need to escape them
all the time in TAPIR KVP requests.

So, the changes included:

* Darwin namespaces.
* TAPIR CNS file.
* All references to Darwin namespaces in output models and response
structures.

I also made a new minor release of TapirLink (0.3.1) that correctly
generates concept identifiers based on the new Darwin schemas.

All TAPIR providers that are using Darwin (core/extensions) as conceptual
schema will need to be reconfigured, regardless the software used.

Please let me know if you find anything wrong or missing.

Best Regards,
--
Renato


> I agree that you need a hash in there - but you may consider a slash.
>
> This is a good account:
>
> http://www.w3.org/TR/swbp-vocab-pub/
>
> It is actually about hosting RDF vocabs but it does cover hash vs
> slash namespaces and it may be worth looking through it before
> finally plumping for hashes just incase there is anything in there.
>
> The ontology is all # based.
>
> All the best,
>
> Roger
>
>
>
> On 18 Apr 2007, at 21:32, Renato De Giovanni wrote:
>
>> Dear all,
>>
>> This is clearly a crosscutting issue and I thought about using the
>> TAPIR mailing list for the following reasons:
>>
>> 1) The main people involved with DarwinCore are subscribed here;
>> 2) This issue raised from a TAPIR use case;
>> 3) It can affect all existing TAPIR/DarwinCore providers, as well as
>> all output models based on DarwinCore.
>>
>> As you know, there was a recent release of TapirLink which includes
>> an LSID authority that serves an RDF representation of DarwinCore by
>> default.
>>
>> Everything seems to be working fine, but when I parse the resulting
>> RDF in the W3C validator, I see that the predicates are being
>> displayed as:
>>
>> http://rs.tdwg.org/dwc/dwcoreGenus
>>
>> While in the semantic world the "expected" representation would be
>> something like:
>>
>> http://rs.tdwg.org/dwc/dwcore#Genus
>>
>> Apparently it seems just a cosmetic thing, but after some quick
>> research this "unexpected representation" can cause problems
>> depending on usage and tools: for instance, if it's necessary to
>> perform RDF/XML round-tripping, then semantic web tools may not work
>> if there's no clear separation between the namespace URI and local
>> names, which is normally done by using the fragment identifier.
>>
>> If you're interested, you can find a similar discussion here:
>> http://www.imc.org/atom-syntax/mail-archive/msg16476.html
>>
>> Which has this interesting follow-up:
>> http://www.imc.org/atom-syntax/mail-archive/msg16480.html
>>
>> Since the new DarwinCore version and its extensions are not yet a
>> TDWG standard and may even be subject to other changes, I'm proposing
>> to add the fragment identifier to all Darwin namespaces. Better to do
>> this as soon as possible if we're going to need this in the future.
>>
>> Please let me know if you have any comments, ideas or concerns...
>>
>> It may be the case that this change will affect other things (like
>> the new GBIF REST service) although probably not as much as
>> TAPIR/DarwinCore providers which will need to re-map their databases.
>>
>> Best Regards,
>> --
>> Renato