Hello all,
I’ve built a mirror that publishes every LCSH heading as a MARC
authority record, rebuilt weekly:
https://bibliotecaalonsocossio.com/lcshmirror/
LC offers bulk LCSH only as linked data (MADS/RDF and SKOS, as JSON-LD
and N-Triples). Bulk MARC is available only through the paid MARC
Distribution Service, which is inconvenient for anyone who needs bulk
MARC LCSH but not necessarily the whole licensed ecosystem.
Each release has the full set (currently 468,352 records) as MARC 21
(UTF-8) and MARCXML, a CSV list (LCCN, 005, heading), and SHA-256
checksums. Files under *-latest names always point to the newest
release, so a cron job can fetch the same URLs every week. There’s
also an Atom feed, and every release is copied to the Internet
Archive.
How it works, in case it’s useful to anyone doing something similar:
- The bulk MADS/RDF dump is used as a census: every heading plus its
recordChangeDate.
- MARCXML for each record is fetched from id.loc.gov one at a time, at
the pace id.loc.gov asks for (one request every 3 seconds). The first
full harvest took 16 days. Weekly runs only fetch what changed, which
takes minutes.
- The weekly change files compare the MARC records themselves with the
previous release, not the dates. This turned out to matter: of 206
headings whose change date moved this week, only 1 actually differed
in MARC. This week: 150 new, 1 changed, 1 deleted.
If you load these next to records from authorities.loc.gov: the 001 is
written without blanks (sh85076432 rather than “sh 85076432 “), so
match on 010 $a. The 005 is the date of the last cataloguing change,
not LC’s August 2025 FOLIO migration, and LC’s internal fields (035,
906/910, 999) aren’t included.
The code is Python (MIT), and the data is public domain (CC0). If
anyone wants to run their own mirror, please do so:
https://github.com/iratedemon04/lcsh-mirror
This is unofficial and not affiliated with LC in any way. Questions,
corrections and pull requests are more than welcome.
Thank you for reading!
Alonso Cossío Vázquez
Biblioteca Alonso Cossío
Received on Sat Nov 07 2026 - 12:35:33 EST