Search

There is something big lurking in your backlog, something that darkens your every work day. Maybe it’s remediation you’ve been putting off for years, or unfinished business from a big migration. Nothing depends on taming the monster, but life would be easier if you did. In our case it was barcodes—tens of thousands of off-site barcodes in the ILS but unavailable to the Aeon requesting system. Every request fulfillment necessitated a tedious manual lookup. Merging the data by hand would have taken untold months of staff hours, and with no common identifiers across systems an automated solution seemed out of reach. By piecing together and adapting tooling from other efforts (Python, Google Sheets, regex), we struck a productive balance between automation and manual work and, while working remotely during COVID-19, added 25K barcodes to our most in-demand collections. The presentation outlines the solution we landed on and makes a limited case for an ad hoc, “bricoleur” approach that eschews sustainability in favor of near-term effectiveness and progressive learning. You may never use your one-off tools again, but you will carry forward something for the next project, and further lower the barrier to taming the next backlog monster.

(Presentation given at the ArchivesSpace Annual Forum, 3 Aug. 2020.) As Columbia University Libraries have progressed from migration to operations in ArchivesSpace, we have developed tools and processes to monitor ongoing work, flag errors and problematic constructs, and report information to archivists and support staff in an automated fashion. The long tail of cleanup after a large-scale migration left a number of known, widespread issues across the corpus, such as missing scope and access notes, incorrectly encoded language information, etc. The regularization required by the migration also opened the opportunity to implement more consistent practices and incremental improvement across repositories in line with DACS principles. To provide continuous visibility and timely warnings, CUL developed a number of automated tools in Python that gather data from both the API and cached EAD files and report information at regular intervals: * Select resource data from entire corpus, reported nightly. * Subjects, agents, and accessions, reported weekly. * EAD checked against custom RelaxNG schema and XSLT rules (formerly Schematron assertions). Status of well-formedness and validity, along with any DACS compliance warnings, reported daily. * ILS catalog sync, reported nightly. The automated reports generate email summaries with links to complete reports in shared Google Sheets. This information helped inform numerous remediation efforts during COVID-19 while all staff are working remotely, such as origination review, access and scope note review, character counts for scope and bio notes to flag collections with limited description.

A large-scale, multi-phase migration of Columbia’s archives into ArchivesSpace was the occasion to brainstorm on flexible, adaptive data analysis and reporting solutions—to think beyond the CSV. Reusable tools built in Python using the Google Sheets API enabled creation of ad hoc, shared reporting interfaces to keep the project on track.