LCCN (was Re: [rdfweb-dev] FOAF and the long dead ...)

Leigh Dodds leigh at ldodds.com
Mon May 9 11:38:29 UTC 2005


conor dowling wrote:

> makes sense except that most people aren't the primary topic of a  
> document - not well recognized documents away. 

Wikipedia entries? If there isn't one, then create one! :)

However, happily you can avoid this by piggy-backing off work
in the Library cataloguing community. (Warning: the following is
the results of some lunchtime digging around I did a few months
ago, but never wrote up properly)

OCLC has some tools that let you look up the Library of Congress 
Authority Records for a given name. These records are specifically
designed to avoid other people have to create "authority control"
systems -- and this is precisely what you, and others are doing.
You want an unambigous IFP independent of email/homepage, etc and
the LC can provide you that.

To continue the example, we can use the LC Name Authority service [1, 2]
to look up "Constantine". There are 368 possible matches, here's a
list of all of them [3]. Note that there's a SOAP interface to this
that allows some degree of automation.

In the list I can see: "Constantius--I,--Emperor of Rome,--d. 306",
(LCCN: n85-121423).

Here's our first candidate for an IFP:

<foaf:Person>
   <eg:lccn>n85-121423</eg:lccn>
</foaf:Person>

There's a human-readable description of the full authority record
linked from the results [4]. It's a persistent URL, so we could be
tempted to also write:

<foaf:Person rdf:about="http://errol.oclc.org/laf/n85-121423.html"/>

I personally don't see a problem with that, but using an eg:lccn
property is a bit more FOAF-y and similar to mbox_sha1sum, etc. You
may also prefer to limit your ties to OCLCs URLs. Another way to
express this might be:

<foaf:Person>
   <eg:authorityRecord 
rdf:resource="http://errol.oclc.org/laf/n85-121423.html"/>
</foaf:Person>

Again this would be an IFP, but here using the authority record
itself in a similar way to foaf:homepage. There's also a stable
MARC-XML version of the data which could possibly be converted to
RDF -- I just tried a MARC XML to DC crosswalk on it, but the
results were pretty poor. The MARC record does include other data that 
may be of interest though, e.g. name variants and in some cases 
publication history.

Basically I think there's some pretty fertile territory here for
working more closely with library/cataloguing communities. They've
spent decades on this kind of stuff so it seems unnecessary to
reinvent all of this.

Cheers,

L.

[1]. http://www.oclc.org/research/researchworks/authority/default.htm
[2]. http://alcme.oclc.org/eprintsUK/index.html
[3]. 
http://alcme.oclc.org/eprintsUK/services/NACOMatch?method=getCompleteSelectedNameAuthority&xsl=http%3A%2F%2Falcme.oclc.org%2FeprintsUK%2FeprintsUK.xsl&serviceType=rest&name=Constantine&maxList=368&isPersonalName=true
[4]. http://errol.oclc.org/laf/n85-121423.html





More information about the foaf-dev mailing list