W3C home > Mailing lists > Public > public-swd-wg@w3.org > December 2007

Re: [SKOS]: [ISSUE 44] BroaderNarrowerSemantics

From: Ed Summers <ehs@pobox.com>
Date: Fri, 14 Dec 2007 13:40:47 -0500
Message-ID: <f032cc060712141040u7e4e1d49m305e85b1e5e70875@mail.gmail.com>
To: "Simon Spero" <ses@unc.edu>
Cc: "public-swd-wg@w3.org" <public-swd-wg@w3.org>, "public-esw-thes@w3.org" <public-esw-thes@w3.org>

Hi Simon:

I've been working on reprsenting LCSH (minimally) as SKOS, so this is
of great interest.

On Dec 14, 2007 12:02 PM, Simon Spero <ses@unc.edu> wrote:
> The problem with the LCSH is that they're  "subject headings disguised
> as a thesaurus" \cite{Dykstra:1988lr}.   Calling associative
> relationships that are not in the least hierarchical  "broader" and
> "narrower" did not prove fruitful.   Weakening the semantics of
> broader/narrower for SKOS would be repeating the same mistake.

I'll check out the Dykstra reference, but there are indeed broader
relationships in LCSH as represented in MARC Authority data [1]. Just
look at the meaning of the 450 when a subfield w has the value g. I've
actually used your data files [2] (thanks by the way!) and created
about 1.5 million SKOS triples.  I've taken a look at some trees of
relationships, and they seem sane for the most part. Here are some
examples (the .svg ought to be viewable in a modern ffx).




The filenames are reflective of the "seed" concept that I used to walk
about a certain radius from. Well and it's Friday... I'd be interested
to hear where you see the broader relationship not working. Also if
you want to generate the triples with your fred2 dataset [2] you could
use the prototype script I've been working on at LC [3]. It requires
python plus the rdflib and pymarc libraries. I'd be very interested in
your feedback on this "naive conversion" as I'm calling it.


[1] http://www.loc.gov/marc/authority/ecadtref.html
[2] http://www.ibiblio.org/fred2.0/authorities/
[3] http://inkdroid.org/svn/lcsh-skos/trunk/lcsh2skos
Received on Friday, 14 December 2007 18:41:01 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:31:46 UTC