Cataloging and metadata: describing collections
How collections are described, in which standards and at what level of detail, and how machine suggestions fit in without being mistaken for a cataloger's work.
A digitized collection is only as findable as its description. An image with no record is lost in a folder, and a record with the wrong date sends researchers to the wrong decade. This section covers how collections are described: in which standards, at what level of detail, and how machine suggestions fit in without being mistaken for a cataloger’s work.
Questions this section answers
- Which standard fits our collection? Dublin Core’s fifteen elements for digital collections, DACS for archival description, EAD for encoded finding aids (the Library of Congress now lists EAD4 as current) and MARC 21 for library catalogs.
- Folder or item? When item-level description is worth the time, and when a good folder description serves researchers better.
- What can AI suggest, and how reliable is it? The Library of Congress’s Exploring Computational Description experiments concluded that “no current AI tools are currently good enough to run fully automatically” for catalog records. The articles look at which fields machines suggest well, which they do not, and how review works.
- How should machine-made fields be marked? Field-level labels that say whether a value was suggested by a machine, reviewed, edited or written by a person.
- How do records get into our system? Mapping fields, vocabularies and identifiers to the system you already use.
Tools and services for these questions
The metadata mapping worksheet lines up your fields with Dublin Core elements, the collection inventory worksheet counts what needs describing, and the file naming builder keeps identifiers consistent between files and records. Our AI cataloging service and AI policy describe review and labeling, and the fictional example collection shows labeled records you can search.
Articles in this section
Each article below answers one question, names the standard it relies on, and labels any fictional example as fictional.
- Issue: Item-level or collection-level description for digitized archivesWhen we digitize a collection, do we need to describe every item, or is a collection-level finding aid with linked images enough?
- Issue: Cataloging photographs with unknown people, places and datesHow do we describe historical photographs when we don't know who is in them, where they were taken or when?
- Issue: Importing digitized records into PastPerfect, CONTENTdm or OmekaHow do we load a vendor's images and metadata into our collection software without breaking records or losing the link between files and catalog entries?
- Issue: How to label AI-generated metadata in catalog recordsWhen AI suggests titles, subjects, names or descriptions for our records, how should we label that metadata, review it and explain it to users?
- Issue: Dublin Core, MARC or EAD: choosing a metadata standardWhich metadata standard should our digitized items be described in, and what does our collection software actually need?
Send one box first.
A pilot box shows you the images, the recognized text and the catalog records before you commit a whole collection. Tell us what is in the box and what you want people to find.
Studio and contacts
24 South Weber Street, Suite 225Colorado Springs, CO 80903(719) 244-8450sales@dastok.com