CONTENTdm, like most platforms that use non-M ARC metadata, doesn’t collate subject headings very well when they are in pre-coordinated strings (e.g., Poultry industry–California–Petaluma–Photographs). This means that when browsing for a topic, photos of poultry operations in Petaluma will be separate from those depicting poultry operations in any other location, as well as from advertisements for poultry equipment. The most common controlled vocabularies for visual objects in the digital library world, such as the Getty Art & Architecture Thesaurus, are either explicitly not applied as strings, or use more limited strings. CONTENTdm’s keyword hyperlinking feature also works against pre-coordinated strings and I want to based the discovery model on faceted subject headings, which would allow the user to narrow a search down by applying a number of limiting terms individually:
Boss of the poultry farm, Petaluma, Cal.
- Topic: Poultry industry
- Geographic: California–Petaluma
- Form/Genre: Photographs
Any number of controlled vocabularies would work for this, but because we have some 35,000 MARC records with Library of Congress Subject Headings
(LCSH) that I will map into CONTENTdm, the choice is already largely set. My solution for making use of the pre-existing LCSH, yet allowing for faceted searching, is to convert LCSH headings into FAST
(Faceted Application of Subject Terminology) headings. Given that many of the geographic, personal and corporate names in the records have not been established in the LC Names Authority File (NAF), only a subset will actually be authorized FAST headings, but the remainder will be in the FAST format and coded as local headings.
Although very few, if any, of the legacy records contain a chronological heading, I have also created FAST chronological headings for all records that provide the decade of the resource, based on the date(s) contained in the MARC fixed fields. This will allow browsing by decade in addition to the topical, geographic and form/genre facets.