[open-bibliography] Library of Congress subject headings & RDF

Fri Jan 7 15:15:45 UTC 2011

On 01/07/2011 07:18 AM, Owen Stephens wrote:
> Can anyone point me at (or advise me on) examples of representing subject
> heading fields from a library catalogue record as RDF. Specifically I'm
> interested in how chained sets of subject headings are represented.

Hmmm... I think of your example as a single heading with multiple parts.
Something like "Popular music -- History -- 20th century" isn't really
about "History" or the "20th century" except as those things relate to
popular music, after all.

The only thing you really lose in the SKOS representation at id.loc.gov
is the explicit typing of the parts.  (That is, there's nothing explicitly
saying that the "20th century" part is a temporal subdivision rather than a
generic topical subdivision.)  Though I've found I can do much of what I need
without tha.  (If anyone here will be at ALA Midwinter on the 10th, I'll
be talking about what we're doing with the LC's SKOS LCSH authorities
in the Subject Analysis Committee meeting Monday afternoon.)

If you really need subfield typing, it can often
be reconstructed in various ways.  Geographic terms follow standard
enough string patterns in LCSH, for instance, that I can generally pick
them out even without explicit typing.  Likewise, temporal subdivisions tend
to have certain patterns, such as '[year]-[year]" or "[n]th century".
Alternatively,  have access to a set of MARC records that has subdivision 
typing, you can analyze the existing strings there and see which strings
tend to go with which type.)

Alternatively, the MADS format that LC will be using in the future is
an RDF format that does include explicit subfield typing in the LCSH
authority data.   It's not yet available on id.loc.gov, but I shared
a podium recently with an LC speaker (Kevin Smith) who said it's on its way.
(SKOS will also continue to be available.)

John Mark Ockerbloom
University of Pennsylvania Libraries