[foaf-dev] alternate names for people

Dan Brickley danbri at danbri.org
Wed Jun 2 13:06:17 CEST 2010

+cc: Yves, who might have Music Ontology perspective on this (single
person acts, vs the person themselves, etc)

On Wed, Jun 2, 2010 at 3:55 AM, Ed Summers <ehs at pobox.com> wrote:
> My apologies if this has come up before. Has anyone needed to model
> alternate names for a foaf:Person before? The issue came up in a
> thread on the openlibrary tech list discussing how to improve access
> to author resources [1].

This is an important use case, thanks for writing it down in our mailing list :)

My first reaction to your proposed design was "oh dear, is this
skos:Concept <-disjoint-> foaf:Person again? Then I actually read the
final W3C SKOS recommendation and was relieved to remember that these
useful label properties have no declared rdfs:domain, ie. that can be
used on anything quite freely. So the SKOS-based design looks
available and usable already.

> Consider Samuel Clemens, who was the author of Huckleberry Finn, and
> whose name in "real life" was Samuel Clemens. Is it reasonable to riff
> on the preferred label pattern in Leigh and Ian's Linked Data Patterns
> [2] and use skos:altLabel here?
> <http://openlibrary.org/authors/OL2119850A/> a foaf:Person ;
>    foaf:name "Samuel Clemens" ;
>    skos:altLabel "Mark Twain" .
> Or is there another well trodden foafy way to do this? Karen Coyle and
> George Oates over at openlibrary.org are seriously looking at using
> foaf to model authors...and they are soliciting ideas on how to get it
> right.

Let's figure this out then.

I think there are two important sub-scenarios here, and the mixing of
bibliography / historical data with social Web data makes their
interconnections very interesting.

Scenario 1: SECRETS. One person has two activities in the world, and
keeps them very distinctly separate. They have an alias or nickname or
alternate persona. Examples here include Fake Steve Jobs
and also http://en.wikipedia.org/wiki/Girl_with_a_One-Track_Mind

On that last example, Wikipedia has "Girl with a One-Track Mind is a
blog by Abby Lee (pen name of Zoe Margolis" ... "Soon after the book
was published, the Sunday Times published an article [3] which
revealed the identity of the author as Zoe Margolis, an assistant
director in the film industry. Margolis described the experience as
"nightmare", "hell" and "fiasco"[1][4], writing about how deeply it
affected her personal life and caused her to lose her career in the
film industry[5][6]".

Scenario 2: common public knowledge of the alternate name, eg.
http://en.wikipedia.org/wiki/George_Orwell ("Eric Arthur Blair (25
June 1903 – 21 January 1950),[3] better known by his pen name George

There is no clear dividing line between the two. Sometimes things
merge slowly as time passes and the pain of revealed identity is
forgotten. But I suggest we need two design notes:

1. To make clear that people have the right to privacy, to keep
secrets, and that just because we know that two personas are the same
real-world person, we don't have to publish that in RDF. An RDF
document containing <foaf:Person foaf:name="Bob"/> and <foaf:Person
foaf:name="Charlie"/> formally says *nothing* about whether there is
one person or two. Many readers and applications will assume it means
there are two people, but it doesn't. Only when we start to use
different-individual or sameAs claims does counting become possible.
In this light, I think describing the personas separately is most
flexible. When such descriptions are linked we end up with two
interchangeable foaf:name properties, "Bob" and "Charlie" and only
contextual metadata to say which one to prefer.

2. When the names are well known to be the same person, and we are
trying to capture uncontroversial common knowledge, then attaching
them in the same person description makes more sense. For this, either
multiple 'name' properties, or the SKOS altName makes sense.

If we go with altName as a recommended pattern, do you think there is
concensus in the bibliographic world as to which of the various names
is the main versus the alternative? Does AACR2 / RDA cover these
rules, for example? Is there something like stageName vs realLifeName
implicit in this?



> //Ed
> [1] http://www.mail-archive.com/ol-tech@archive.org/msg00048.html
> [2] http://patterns.dataincubator.org/book/preferred-label.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