Re: Trailing slashes in RDF

My hack I am afraid - I was trying to be good and distinguish the
person and account without having to mint #account or something that
does not appear in the HTML version.


..but it is not ideal as for some reason (I didn't investigate
further) you need to be logged in (As anyone) to retrieve the slash
version at ORCID - http://orcid.org/0000-0003-0782-2704/

Note that ORCID does not elsewhere use the trailing-slash URI - I
didn't want to modify the HTML bit as it is essential that people who
use ORCID identifiers use the proper identifier for the person, and
not the profile (thus
without the /)


See https://github.com/ORCID/ORCID-Source/blob/master/orcid-api-common/src/main/java/org/orcid/api/common/writer/rdf/RDFMessageBodyWriter.java
to suggest improvements - perhaps reverting it to something like
#profile?


For ORCIDs I agreed early with the ORCID developers that the URI
http://orcid.org/0000-0003-0782-2704 is the identifier, and that it
defines the foaf:Person and not his ORCID profile -- so the old #me
trick was not work here.


I still think the /-pattern can be useful elsewhere - as it would work
even on pure HTTP servers with folders, which would do the redirect
for you (But with 302 instead of 303).


Another hack (which will blow up the Python's URI parser) is to use
the empty #fragment to identify the real object:

<http://example.com/object> foaf:primaryTopic <http://example.com/object#> .



On 6 March 2015 at 17:25, Phil Archer <phil@philarcher.org> wrote:
> Dan,
>
> IIRC you asked a while back on this list about the semantics associated with
> two URIs that differ only in the presence/absence of a trailing slash. I've
> been spending some time looking at the RDF available for ORCIDs recently,
> for example [1]. It's difficult to spot but in fact they use a URI like
> http://orcid.org/0000-0003-0782-2704 to refer to a person (in this case Ivan
> Herman) and the same URI with a trailing slash to refer to his online
> account.
>
> Clearly these two URIs *are* different and are being used consistently to
> distinguish between a real world object and its description, but, by
> convention, do you think the difference is sufficient to be regarded as
> 'safe?' Is orcid.org following good practice here?
>
> FWIW, no, one URI does not redirect to the other.
>
> WDYT?
>
> Phil.
>
> [1]
> http://i-sieve.com/cgi-bin/HTTP_Headers.cgi?url=http%3A%2F%2Forcid.org%2F0000-0003-0782-2704&UA=ttl&get=on
>
> --
>
> Phil Archer
> http://philarcher.org/
> +44 (0)7887 767755
> @philarcher1
>



-- 
Stian Soiland-Reyes, eScience Lab
School of Computer Science
The University of Manchester
http://soiland-reyes.com/stian/work/    http://orcid.org/0000-0001-9842-9718

Received on Tuesday, 10 March 2015 11:30:04 UTC