On 20/12/2013 15:01, Paul Shackleton wrote:
> I would point you in the direction of OCLC's Virtual International Authority File (VIAF), Librarians have been dealing with defining individual names for a very long time. See the following links for the interface and info on the API and research:
>
> http://www.viaf.org/
>
> http://www.oclc.org/viaf.en.html
>
> http://www.oclc.org/developer/services/virtual-international-authority-file-viaf
>
> http://www.oclc.org/research/activities/viaf.html
Paul,
I'm aware of VIAF, and it raises some interesting issues about the
problem of identifying people unambiguously.
VIAF is an "after the event" reconciliation service, which brings
together name authority data from participating institutions (mostly,
but not all, national libraries). This relies on the people in question
coming to the notice of bibliographic cataloguers, either as authors or
(presumably) as subjects within published works. The name authority
entries themselves found in VIAF are far from unique (there are five
people other than me with the name authority "Light, Richard"), and one
thing which VIAF does right is to assign a unique, numerical identifier
to each cluster of records signifying the same individual. Thus I have
the VIAF identity www.viaf.org/viaf/67713682. This identifier is
amenable to Linked Data principles, so I can use it to grab my data as
RDF, XML, JSON, etc. Thus far, all is lovely.
One problem I have with VIAF is that I don't see how I would use it to
disambiguate an individual programmatically. If I run a search for
"Richard Light", I get an HTML page listing 17 hits, and I have to scan
it by eye, with a knowledge of what I have published, to work out which
one is me. I then have to follow that link, again manually, to find out
what my VIAF identity is. However, looking more closely at the VIAF
site, I find that there is an OpenSearch interface [1] which allows me
to put my search in a way that returns an XML response [2]. This is
useful; for example one could set up a "web termlist" to return these 17
records, and let the user pick which one matches the person they are
cataloguing. (This is what I have done, for example, with Geonames.)
However, the problem remains that the name authority information against
which one is searching is not designed for this purpose, and so is
suboptimal. It is a matter of chance which of a person's names and
initials will be used in a name authority record, and whether their
dates of birth and death will be included in it. So you have no idea
what to search for. Typically dates will not be given, and /place /of
birth and death doesn't feature in the VIAF framework at all. (There's a
Richard Light in the list with the same birth year as me, who isn't me ...)
The other problem with VIAF is that it is designed to be populated by
national bibliographic agencies with information relating to published
works. Thus (a) it isn't open to "crowd sourcing" approaches and (b) by
design, it can only deal with a small subset of the human race.
Thanks for the prompt: it led me to discover that you can use the "web
termlist" approach with VIAF, something I hadn't appreciated.
Best wishes,
Richard
[1]
http://oclc.org/developer/documentation/virtual-international-authority-file-viaf/request-types
[2]
http://viaf.org/viaf/search?query=local.personalNames+all+%22Richard%20Light%22+&maximumRecords=100&sortKeys=holdingscount&httpAccept=text/xml
--
*Richard Light*
****************************************************************
website: http://museumscomputergroup.org.uk/
Twitter: http://www.twitter.com/ukmcg
Facebook: http://www.facebook.com/museumscomputergroup
[un]subscribe: http://museumscomputergroup.org.uk/email-list/
****************************************************************
|