[foaf-dev] alternate names for people

Dan Brickley danbri at danbri.org
Wed Jun 2 16:30:25 CEST 2010


On Wed, Jun 2, 2010 at 3:59 PM, Ed Summers <ehs at pobox.com> wrote:
> On Wed, Jun 2, 2010 at 9:28 AM, Houghton,Andrew <houghtoa at oclc.org> wrote:
>> Multiple names are all variants.  When an "agent" chooses one it
>> becomes the preferred name *for that agent*.  For example, the
>> Library of Congress, an agent, may choose a preferred name from
>> the list of variants, but the German National Library, an agent,
>> may choose a different name from the list of variants.  Neither
>> agent is wrong, that is why it's *preferred*.  So with *preference*
>> there is some provenance that needs to be attached.
>
> In skos this provenance for a label is normally limited to the
> language, and the context of being attached to a particular concept in
> a particular concept scheme. If you need to be able to annotate labels
> more there is always the skos-xl extension [1].

I'd forgotten about xl. Can you help us understand
http://www.w3.org/TR/skos-reference/skos-xl.html#Label ?
http://www.w3.org/TR/skos-reference/skos-xl.html#literalForm lists
literal form as a 'class' but also gives it a domain. Confused! A full
example would be sweet...

> So it sounds like there might be some preference (ahem) for modeling
> Samuel Clemens and Mark Twain as distinct resources of type
> foaf:Person, and then linking them together with some property?

"Distinct resources" is slippery. The class foaf:Person is pretty
inclusive, and covers living/dead, real/imaginary etc. The further you
get from human, the less applicable it is.
But we have to separate "distinct descriptions" from "distinct
things". One one perspective, Samuel Clemens simply is Mark Twain. So
owl:sameAs would be quite proper. They share the same birth
certificate, for example (regardless of what it says on it); which via
other routes could imply common identity. If there were any people out
there with Samuel Clemens of Mark Twain as biological fathers, those
people (short of scifi) only had one biological father, etc.

> I agree with Reto that owl:sameAs seems kind of strong here. Perhaps the
> Linked Data community isn't too concerned with the logical
> implications of owl:sameAs though?

owl:sameAs means what it means, and we have to respect that; you use
it when there's only one thing, but somehow different mentions of it.

For actors and musicians and a few other cases, we could perhaps
elevate one of someone's personas to distinct fictional personhood,
and use other relationship types to make clear ('aka' etc). But I
think we'll need (perhaps per application/project/dataset, sure) to
pick one of their persons as the real/main/hub one.

> I'm still not entirely convinced that this isn't a good use for
> skos:altLabel or skosxl:altLabel.

It could well be. At some point, since reification was mentioned,
we'll go into the rats nest of
dealing with changes over time. The ability to say of a person, "this
is a description of them in some [typically temporal] context" is
quite useful. That would cover changes of name, of gender, employer
and age in one go. RDF99's reification design is not up to this job,
but attaching metadata to descriptions isn't rocket science either,
and SPARQL offers some support.

Say I change my name to Stan, we can still annoted
danbri-foaf-2000-06-30.rdf as "a description of dan brickley, asserted
true in June 2000". We could even SPARQL query to ask how I described
my name or xyz:bloodType prior to 2010, without having to come up with
a vocab design for each personal characteristic. This strikes me as
simpler than having this conversation once for names, and again for
each other kind of potentially contextualisable property. But it does
involve living outside the simple worldview of triples, and embracing
a quads / contexts / factoids perspective.

I think I'm happy doing that, since we have altLabel (and xl:
version?) for the simplest form, we have the option to describe people
in 2 distinct descriptions as a 2nd more complex form, and to assert
either owl:sameAs or perhaps foaf:aka between those mentions.

For anything more complex, I think we need to get our noses out of the
triples and use quads. The question then becomes, what other kinds of
context other than time might be enumerated. There are names used
professionally, there are names used before / after marriage, or
before/after other forms of name-change. There might be stuff in
http://www.w3.org/2005/Incubator/prov/wiki/W3C_Provenance_Incubator_Group_Wiki
that's relevant here...

cheers,

Dan

> //Ed
>
> [1] http://www.w3.org/TR/skos-reference/skos-xl.html
> _______________________________________________
> foaf-dev mailing list
> foaf-dev at lists.foaf-project.org
> http://lists.foaf-project.org/mailman/listinfo/foaf-dev
>


More information about the foaf-dev mailing list