Data models in ArchivesSpace and the Digital Collections

This overview of the data models in ArchivesSpace and the Digital Collections should make future conversations about these two websites easier.

Tabular vs. hierarchical metadata

The Digital collection is primarily a table of works, each with complete metadata. Each work has one or more assets, which are computer files stored in S3.

The metadata in ArchivesSpace is hierarchical, not tabular. Each level of description is understood to inherit all the metadata from levels above it, except where that metadata is explicitly overridden.

ArchivesSpace’s primary function is to provide intellectual control over archival collections, but it also keeps track of the physical locations of folders, boxes and so on. (e.g.: inventory of a physical box; inventory of an archival (intellectual) series).

Links between the two sites

The Digital Collections has pointers to both physical locations (boxes and folders, e.g.) and more abstract archival entities like series and sub-series (and of course collections).

ArchivesSpace has many pointers to individual works in the Digital Collections, but these are not currently accessible to the public.

Let’s take a look at a letter from the Beckman collection as an example.

Digital collections:

In the digital collections, the letter takes the form of a work: https://digital.sciencehistory.org/admin/works/wm117p03j

To place the work in the context of the collection’s archival arrangement, the D.C. gives you the following clues:

Collection

• The letter is part of a collection, the Beckman Collection.

  • A work can be part of more than one collection, but

  • A collection cannot be part of another collection.

• The work is part of a sub-series and a series within the Beckman collection.

Series arrangement

In the digital collections, series and sub-series arrangement is stored as an unordered sequence of strings attached to the work. In this case we have:

• Series Arrangement

  • Series I. Arnold O. Beckman Files

  • Sub-series 1. Correspondence

• Each string concatenates the type of metadata ( Sub-series), the identifier, (I.), and the title of the grouping: (Arnold O. Beckman Files). These are stored separately in ArchivesSpace.

• The d.c. does not encode the fact that the series contains the sub-series, rather than vice versa.

• Likewise, there’s no data in the digital collection that would allow you to order the sub-series within a given series, or the series within a collection.

ASpace Reference Number

The work also has an ASpace Reference Number: 118f36c4c5a373e4b4a81253ebc85fae.

• This ASpace Reference number can tie a work or collection in the D.C. to a file, sub-series or series in ArchivesSpace - any description level that is an archival object.

• In this case the Reference Number refers to a file in ArchivesSpace (see below).

Physical Location

• A work’s physical location is encoded as set of seven key-value pairs (the keys beingbox; folder; page; part; volume; shelfmark; and reel). Archival records in the D.C. so far have only used box, folder, and reel. (The others are in use to catalog rare books and museums items).

• For this work, all the keys except box and folder are blank.

  • box is the string 1; folder is the string 29.

ArchivesSpace:

Digital object

In ArchivesSpace, the letter takes the form of a digital object.

• ArchivesSpace maintains a distinction between a digital object and an archival object.

• Like all digital objects, it has been unpublished since 2022.

• Title is the same as the D.C. work title.

• Metadata contains a link to the work in the digital collections. (The work does not have a link back to the digital object, but the work does have a link to the file the digital object is part of.

• URL: https://sciencehistory.libraryhost.com/admin/digital_objects/247#tree::digital_object_247

• Digital objects were not part of the earliest versions of ArchivesSpace (item-level description is uncommon in archival practice as it’s unsustainable at scale).

• The digital object is part of a file.

Item

For completeness' sake, note that we do (rarely) describe individual physical objects (e.g. this film reel) as items.

• Items are archival objects and thus have Ref IDs. (Technically any archival object can contain another archival object, but in the case of an item this would be very unlikely.)

File

• In this case, the file is a digital surrogate for a manila folder (folder 29 in box 1) which contains the letter.

  • Nothing to do with a file in the operating system sense.

• Title: Los Angeles Chamber of Commerce - Air Pollution Committee, 1951-1954

• URL: https://archives.sciencehistory.org/repositories/3/archival_objects/10615

• Is an archival object, as opposed to a digital object.

  • In theory, any archival object can contain another archival object.

  • In practice, at least for the Beckman collection, the file is the lowest level of archival description above the digital object.

• Items, files, sub-series and series are all considered archival objects. Digital objects and collections are not.

• All archival objects have a unique ID called a Ref ID.

• Ref ID:118f36c4c5a373e4b4a81253ebc85fae.

Sub-series

• URL: https://archives.sciencehistory.org/repositories/3/archival_objects/5

• Is an archival object.

• Ref ID: 66a590971707f99df33fc42be0d0c909

Series

• URL: https://archives.sciencehistory.org/repositories/3/archival_objects/1

• Is an archival object.

• Ref ID: 5575406909262fd92cf89083a49f855b

Collection

• URL: https://archives.sciencehistory.org/repositories/3/resources/1

• The collection is not an archival object, but a resource.

  • Hence, it does not (and cannot) have a Ref ID.

• Has an accession number: 2012-002

  • Accession numbers are arbitrary strings and might contain digits, spaces, letters and punctuation.

• Has an internal ID, like all resources: in this case the integer 1, which is at the end of its URL.

• We use the ID as part of the file name at the EAD export page.