LUSTRE event: AI & born digital archives

Adam Nix and Stephanie Decker recently took part in a fascinating workshop on digital archives at the Cabinet Office in London, organized by the fantastic LUSTRE network. The overall aim of the LUSTRE project is to connect policymakers with Computer Scientists, Digital Humanists and professionals in the GLAM sector (Galleries, Libraries, Archives and Museums). The project is co-delivered with professionals from the Cabinet Office’s Central Digital and Data Office (CDDO). The recordings from the day are available here.

We talked about our recent paper in AI & Society: Finding light in dark archives: Using AI to connect context and content in email. The practice of digital archival discovery is still emerging, and the approaches future research will take when using digital sources remain unclear. Archival practice has been shaped by paper-based, pre-digital sources and guides assumptions around how researchers will access and make use of such collections. Paradoxically, dealing with the increasing relevance of born-digital records is not helped by the fact that many born-digital collections remain dark, in part while questions of how they should be effectively made available are answered. Our research takes a user perspective on discovery within born-digital archives and seeks to promote more meaningful access to born-digital archives for researchers. In doing so, our work deals with the implications that unfamiliar archival technologies (including artificial intelligence) have on disciplinary traditions in the humanities and social science, with a specific focus on historical and qualitative approaches.

Our work in this area currently focuses on the issue of context within organisational email, and the challenges of searching and interpreting large bodies of email data. We are particularly interested in how effective machine-assisted search and multiple pathways for discovery can be used to open contextually opaque collections. Such access is likely to leverage a collection’s structural and content characteristics, as well as targeted archival selection and categorisation. We ultimately suggest that by combining relatively open user-led interfaces with pre-selective material, digital archives can provide environments suited to both the translation of existing research practices and the integration of more novel opportunities for discovery. Our presentation will summarise our progress in this area and reflect on the technical and methodological questions our work here has raised.