The Rockefeller Archive Center has a long relationship with Archivematica; I think we were one of the first institutions to implement Archivematica in production. Our use of Archivematica coincides with our use of PREMIS Rights. Early on, we decided that we would always add PREMIS Rights statements to transfers to support mediated access to digitized and born-digital records. Having machine-actionable rights statements allows us to flexibly develop and change our implementation of mediated access without being tied to a particular model or definition.

As part of scaling up our digitization and born-digital transfer processes, we wanted to distribute the work of authoring and managing PREMIS rights statements across the organization. However, particularly when serialized in XML, a significant degree of specialized expertise to understand and author PREMIS Rights. Over the years we’ve created several different kinds of graphical user interfaces aimed at lowering the barrier to the human management of PREMIS Rights.

Archivematica PREMIS UI

Archivematica provides a built-in UI for authoring PREMIS Rights statements, and that’s how we started out associating rights with ingests. However, we ran into some areas of friction:

  • The form for creating a rights statement is broken into two pages, one for the basis and one for the acts. This requires users to remember what they entered in a previous page. There are also just a lot of fields.
  • Second, this UI only allows you to associate a single set of rights statements with a transfer. That means all the objects in that transfer need to have the same rights, which is not always the case. This resulted in breaking up transfers by rights instead of more archival concepts like provenance.

CSV Import: object-level PREMIS Rights import

Our first attempt at moving the interface for creating PREMIS Rights outside of Archivematica was specifying and sponsoring a new feature which supported the import of PREMIS Rights via a CSV file. This feature also allowed these rights to be specified on a per-object basis. While this feature simplified the serialization of PREMIS and supported some degree of automation, creation of these rights statements (and the management of the scripts necessary to create the spreadsheets) still required a lot of specialized knowledge, so we knew we’d have to look for other solutions.

Aurora: PREMIS Rights from donor agreements

A few years ago, we developed a bunch of different applications and systems integrations to support the ongoing transfer of born-digital records from our donor institutions. The “front door” to this pipeline is an application called Aurora, which supports the initial transfer, validation, and appraisal of packages of records.

One of the things this application does is automatically assign PREMIS Rights to records packages by taking the basics of a rights statement and calculating dates based on metadata associated with the records package. This means that every born-digital record that we have custody of has at least one machine-actionable rights statement associated with it. It’s also important to say that these rights statements are translations of prose in our donor agreements. I’ll touch more on this later, but that process of translation has proven to be challenging.

Aurora allows users to create, update and view rights statements. Because it is a Django web application just like Archivematica, we ended up “borrowing” much of the code for our implementation from Archivematica and then making some enhancements.

  • The first major change we made was moving all the elements of a rights statement (basis and acts) to the same page. This makes it possible for users to holistically understand the rights statement while they are creating it.
  • We also added a human-readable display of rights statements, to build confidence for archivists and donors that the correct rights had been assigned to a records package.

We went through several rounds of usability testing and were able to fix several issues. But one of the things we were unable to resolve was the translation challenge I alluded to earlier. This is less of a technical issue, or a problem with PREMIS and more about the specificity of language in donor agreements are as well as the kinds of judgement necessary to interpret often purposefully fuzzy prose into data elements.

Aquila: PREMIS Rights as-a-service

In our next and most recent iteration, we took a broader view. Taking into consideration digitized as well as born-digital records, we started by thinking through our organizational processes and policies around the creation and management of rights and clarifying two concepts which we then modeled in the application we developed.

  • The first of these is a Rights Shell which is basically a PREMIS rights statement with rules for dynamically calculating begin and end dates.
  • We also formalized the idea of a Records Grouping, which is carefully worded to distinguish it from both a record group and a collection. These groupings are not necessarily determined by provenance or internal structure of the records but are more things like “Online exhibit content” or “personal papers” which often cut across collections or other aggregations of records.

Once we’d formalized these concepts, we then worked with our head of arrangement and description to create nine standard rights statements that cover most of the cases present in our holdings. We also developed a standardized and documented way of adding new rights statements which ensures the proper organizational guardrails are in place for this work going forward.

This conceptual and procedural foundation allowed us to create a much simpler UI, particularly for date rules. We were able to break date rules into some broad categories, and then control the visible fields in the UI based on that category. This tweak was a huge usability improvement. Showing the user only the necessary fields removes a ton of cognitive overhead.

Aquila also has a REST API which supports integration with other systems and services, including an endpoint which can return a fully valid PREMIS rights statement given a package date range and an identifier for a Rights Shell. We currently use Aquila in our pipeline for ingesting digitized content into Archivematica, and plan to use it in other similar pipelines for digitized AV and born-digital records.

Conclusion

One of the main things we learned during the years of working on this problem is that rights management is not just a technical problem. PREMIS gives us a great structured framework for rights, but our institutional ways of thinking about rights are often fuzzier and less formal than PREMIS wants us to be. People and machines are good at different kinds of things, and enabling people to create machine-actionable rights statements requires a lot of forwards and backwards translation. These conditions mean we must take an iterative approach that’s based on evidence gathered from usability testing.

I wanted to offer a huge thank you to Bonnie Gordon, whose thinking and doing was instrumental to all our efforts around PREMIS Rights. Bonnie is an incredibly sharp thinker who can cut through complex concepts to provide clarity which instigates institutional change. Although Bonnie has moved on from the RAC, we’re grateful for her work every day!

This blog post was presented as a lightning talk at the PREMIS Implementation Fair at iPRES 2023.