[rdfweb-dev] advocating use of rdf:ID
zednenem at psualum.com
Tue Aug 19 04:41:17 UTC 2003
Libby Miller writes:
> I think the key issue is that a thing can't have two urls in RDF, but
> a foaf:Person could have two foaf:mboxes; so if someone uses the
> 'wrong' uri for you (and many people use many uris for the same
> person in their foaf files) - then RDF will see the different yous as
> distinct objects. Only smushing could make them the same.
> However, if your email address changes and you want people to use the
> new one, the migration path is simple: you put both in your foaf
> file, and smushers should then combine all the subjects of both
> addresses. What you *can't* say in FOAF is that both these uris refer
> to you; also you can't say, please smush these things together,
> because only owl:inverseFunctionalProperties can do that as it
> stands, and the uri for an RDF resource isn't a property of that
> resource in that sense.
Resources are resources, whether they're identified by URI or not.
You'll still need to smush them together based on foaf:mbox,
foaf:mbox_sha1sum, and foaf:homepage.
If I get a new e-mail address, I just need to have these triples in my
file and any agent can tell they both refer to the same person:
_:x foaf:mbox <mailto:me at example.org>.
_:x foaf:mbox <mailto:also_me at example.org>.
_:y foaf:mbox <mailto:me at example.org>.
_:z foaf:mbox <mailto:also_me at example.org>.
Any useful FOAF agent will be able to determine that _:x, _:y, and _:z
are the same person. The same process that lets it determine that would
also tell it that the subjects of these triples are the same:
<http://example.org/foaf#me> foaf:mbox <mailto:me at example.org>.
<http://also.example.org/me> foaf:mbox <mailto:me at example.org>.
RDF and FOAF are completely agnostic about whether you identify a person
with a URI or not. However, I personally would recommend against using
an HTTP URI to identify someone unless you got some information about
that person when it was accessed (preferably with the response including
a Content-Location: header). For that reason, I would recommend
*against* using rdf:ID, because that associates a fragment identifier
with a person, which is unlikely to be useful in a web browser. But
that's just my opinion.
David Menendez <zednenem at psualum.com> | "In this house, we obey the laws
<http://www.eyrie.org/~zednenem> | of thermodynamics!"
More information about the foaf-dev