Given the recent discussion of web search engines, this seems a good
time to return to the issue of controlled access terms in EAD encoded
documents, and the relationship of MARC records to EADd inventories.
I put a query out on the web recently and got the following
responses (summarized). I apologize if I'm misrepresenting anyone here.
I got two answers from Harvard. One response indicates that the
Schlessinger routinely transfers the controlled language from the MARC
record to the EAD document in a <controlaccess> section, but has not
decided about subjects, geonames, etc. in unit titles. They have a
collection of genre terms they use routinely. They've not decided about
subject (topic) terms. They are still working on the full-name-to-initials
issue (i.e., where we write out the full name once and then substitute
initials thereafter) and have thought of creating the full names as
entities so search engines pick them up. Currently they're marking up
"prominent subjects/geonames (in S+C and unittitles) is specific
Massachusetts towns," but acknowledge that subject cataloging is highly
subjective. They also recognise that web search engines are pushing the
power of natural-language searching and wonder how long controlled topical
terms will survive.
I'm inserting the second Harvard response, unsummarized, from
> name authority
(Apparently) unusually, Houghton has always done name authority work for
all names of authors and recipients. We continue to do so. The only
thing we are dropping has to do with birth and death dates, when we're
marking up a recipient, e.g.
Victoria. </persname>Letters to <persname type="LCNAF">Virginia Wolff,
</unittitle><unitdate type="inclusive"> 1914-1916. </unitdate></did></c>
to have birth and death dates in line was easily confused with inclusive
dates of the correspondence.
> implicit concepts not explicitly stated
These go in <controlaccess> according to the tag library. I think this is
fine as far as controlled subject headings are concerned, but otherwise I
believe in following the general MARC principal that you don't make an
added entry unless it's apparent from the record why it's there.
> links between the MARC record and the electronic finding aid,
We require that every finding aid have a corresponding MARC record, and be
linked to it (although since we don't yet have a Web catalogue, this isn't
> subject headings of all sorts (topic, occupation, genre)
I've been limiting subject headings (LCSH) and occupation (I can't
remember the formal name of the thesaurus, but I think it was done by
RBMS) . . . to the MARC collection-level record. We traditionally do very
little subject cataloging, and I'm not inclined to change that. We use
AAT for genre terms, and have been using controlled vocabulary in
constructing folder descriptions for five years now.
> If you're adding them to the finding aid, where do you put them?
We mark them up in context. We don't cluster controlled vocabulary terms
together in a finding aid--that's the function of the MARC record.
Schlesinger Library does, however, if you want to look at one of their
on-line finding aids.
Metropolitan Toronto Archives does not use MARC records or EAD, but does
apply a high level of thesaurus-based vocabulary control to the records
they create and display through a Gencat relational database, with a
search engine that does some vocabulary mapping from natural language to
UVa. has started discussing adding controlled vocabulary, but
hasn't gone beyond the talking stage. They have considered adding the OPAC
"tracings" to the EAD document. The cataloging department has suggested it
could develop a list of commonly used names in their collections to which
the catalogers would then add the LCNAF entry which the recon people would
then add to the guides. The catalogers would not develop "authorized"
names for those names that appear infrequently.
NC State uses LCNAF and LCSH and plan to move to more specific
thesauri for only those areas they have identified as specialties.
UC Berkeley checks personal and corporate names in corresponence
lists against LCNAF. Since they find that many topical subjects and
thesaurus terms founf in the catalog record show up as keywords in scope
notes, they don't enter them.
Elizabeth H. Dow, Ph.D. [log in to unmask]
Special Collections -- Bailey/Howe Library
University of Vermont