Implementing ArchivesSpace at NYPL: Part 2
Since publishing our first post, the Archives Unit has successfully implemented ArchivesSpace as a data management system in daily operations. Our implementation process was guided by the realities of day-to-day processing operations, best practices and standards, and the capabilities of our applications.
Here's a refresher of our four guidelines:
- How can we adapt the tool to meet our existing descriptive practices?
- How ought we adapt our local practices to meet ArchivesSpace’s expectations?
- How can we modify the application to expedite minimal processing?
- How can we integrate ArchivesSpace into our metadata ecosystem?
In this follow-up post, we'll explore those final two questions that guided our implementation.
Adapting ArchivesSpace for processing
The Archives Unit prides itself on applying a "More Product, Less Process" (MPLP) approach to archival processing, which applies broad minimal description to quickly provide access to collections.
During our evaluation of ArchivesSpace, we found that its maximalist approach to data entry was a barrier to using MPLP-style processing. ArchivesSpace provides fields that go into far deeper detail than NYPL would provide for description; while these allow for a degree of descriptive power, very few of these fields are used in an MPLP environment, and result in a cluttered interface.
Here, you can compare the difference between the default ArchivesSpace interface and a modified version for The NYPL, and see further details below.
Our solution was to use ArchivesSpace’s built-in features to provide default values for commonly used fields. For example, because the Language of Description of our finding aids will almost always be English, this can be set as a default value. We also automated the generation of many descriptive elements, thus further expediting processing. In particular, we developed and implemented an automatic date parser, which generates machine-readable date metadata with no special input from archivists.
In cases where a data entry field had a compelling but uncommon use case, such as Sponsor Notes and Physical Dimensions, we kept the field in the staff interface but documented its use (or lack thereof) in our local manual.
In addition to processing via the ArchivesSpace staff interface, the Archives Unit developed a spreadsheet-based workflow for processing larger collections. This allows for minimal basic data entry as well as flexibility to shift/rename/rebox components easily during processing. While a spreadsheet will not contain comprehensive description (e.g., there are no fields for container profiles or collection-level metadata), the basic description can be enhanced within ArchivesSpace post-import.
One of ArchivesSpace’s primary advantages over a FileMaker-based system is its ability to easily exchange data with other systems. As a web-based application with an accessible API, ArchivesSpace can interact with our archival systems.
Instead of doing a single large import of our existing finding aids into ArchivesSpace, the Archives Unit decided to bring collections in as needed. To expedite this process, and to provide fine metadata controls, we developed a collection-level import pipeline from our Archives Portal to ArchivesSpace. This integration allows archivists to import collections into ArchivesSpace for enhancement or additional material. While a direct integration between ArchivesSpace and the Archives Portal is not yet in place, we plan on adding this feature during 2018 to easily make finding aids accessible.
To support our audiovisual digitization initiative, we also developed an integration pipeline for our Metadata Management System, which manages representations of digitized material. For the digitization initiative, archivists will link digital files to their descriptions within finding aids for publication. Because this linking will happen in ArchivesSpace, we needed to develop a data pipeline for digital asset metadata from our MMS into ArchivesSpace. To resolve this, we created an export function in our MMS that will create Digital Objects in ArchivesSpace for digital audiovisual material, which can then be linked to components.
In all, our implementation of ArchivesSpace has been successful—we achieved our goal of rolling out the application while maintaining existing workflows and integrations, and took advantage of new opportunities to improve our processes and develop new pipelines between systems. We look forward to continuing our work with ArchivesSpace through 2018 and beyond.