The digital transformation is turning archives, both old and new, into data. As a consequence, automation in the form of artificial intelligence techniques is increasingly applied both to scale traditional recordkeeping activities, and to experiment with novel ways to capture, organise, and access records. We survey recent developments at the intersection of Artificial Intelligence and archival thinking and practice. Our overview of this growing body of literature is organised through the lenses of the Records Continuum model. We find four broad themes in the literature on archives and artificial intelligence: theoretical and professional considerations, the automation of recordkeeping processes, organising and accessing archives, and novel forms of digital archives.
AI usage in Digital Archives
AI, in this sense, is used to bring alternative, more fluid orderings to the archive, for example, by indexing it using the contents of the records. While archival principles discover their limitations, they also could find new life in a datafied world.
Metadata
Another area of recordkeeping processes in which AI is discussed focuses on issues of description and extracting metadata. Büttner [2019] reports on a small but successful proof of concept project in which semantic rule-based auto-classification for official documents in the Council of Europe was developed to relieve record creators from providing metadata and increases the quality of subject metadata. Spencer [2017] describes how automated methods can be applied to identify relationships between records and how these technologies may help provide additional contexts to records.
The increasing amount of digitised and born-digital archival materials calls for the automation and even re-invention of the organisation and access to records. Several related contributions fall within the “Organise” and “Pluralise” Records Continuum dimensions, in the sense that they assume the existence of records that have already been created and captured. Yet, in so doing, they often create novel archival information in turn. A common theme is that AI can be used to move beyond the traditional principles of organising and accessing archives by provenance and original order, which focused on inventories and descriptions of archival units. This approach can now be complemented by other means, primarily focused on the contents of records.
The use of AI to re-think how archives can be organised starts with the automatic extraction of contents and their indexing. While our focus here is not on technical document processing—see Binmakhashen [2020] for a recent overview—we single out the contribution of Transkribus in this respect [Muehlberger 2019]. Transkribus is a tool that integrates image and text recognition models to facilitate Handwritten and Optical Character Recognition. The tool, accessible via a graphical interface or programmatically through APIs, is designed to “fit neatly in the archival workflow, making direct use of growing repositories of digitised images of historical texts.”
Searching and retrieving information contained in archival records has also been an active area of work. Lee [2019] uses visual template matching and automatic classification to retrieve reference cards from the International Tracing Service digital archive, one of the largest and most heterogeneous collections of Holocaust-related material. The need to aid archivists and other users to sift through possibly relevant records via automatic classification is also addressed by other predictive frameworks [Risi 2019]. Bell [2020] takes, instead, a different approach based on archival catalogs. This work proposes to automatically aggregate archival description using text-mining techniques to aid their exploration.
By submitting this form, you agree to allow SilkWay Development to collect, store, process and use your information and data, and you agree to receive email communications from SilkWay Development.