The Archives and Records Association UK & Ireland (ARA) has awarded two grants to projects investigating AI and their application within the recordkeeping sector.
The Archives and Records Association UK & Ireland (ARA) has awarded two grants of £10,000 to projects investigating AI (and other emerging technologies) and their application within the recordkeeping sector.
The call for proposals was made in April 2024 and had a very loose brief, inviting proposers to make bids for work simply within the remit: AI and Emerging Technologies
The grants have been awarded to:
Dr Lise Jaillant (Loughborough University, UK) and Dr Giovanni Colavizza (Odoma LLC and University of Copenhagen, Denmark) with main partner The National Library of Wales and with support from The National Archives (TNA) for a project entitled: FLAME (AI For Libraries, Archives and Museums)
And
Dr Alexandrina Buchanan and Dr Victoria Stobo of Liverpool University Centre for Archive Studies (LUCAS) in collaboration with the Liverpool University Digital Innovation Facility for a project entitled: AI and ML for Catalogue Conversion.
In making these grants the Archives and Records Association acknowledges both the potential benefits and the potential risks associated with using AI in the recordkeeping sector.
John Chambers, Chief Executive of the Archives and Records Association said:
“AI is now real and is a fast-moving issue for the record keeping sector. These grants were deliberately targeted so that ARA can help make advice and guidance available to those who need it as soon as realistically possible. The results of these projects will be made freely available to everyone to use to support their work. We expect to be involved in AI research with a focus on the practical use of it over the coming years. This is a start.”
The outputs from these projects should help to make a difference to the sector and to the wider Galleries, Libraries, Archives and Museums (GLAM) sector.
In the recordkeeping, and wider GLAM, sector the purpose of collections and the preservation of documents and objects is to make them available (whether immediately and publicly, or at a future point, or with some restriction); they are meant to be used, not locked away. However, digital collections are often inaccessible, due to issues with the lack of data preparedness, data protection, clarity on copyright, and the need to deal appropriately with sensitive contents.
The vast amounts of data within digital collections present a difficulty. The analysis cannot be done manually (it would take hundreds of years) and therefore automation is a necessity and Artificial Intelligence (AI) and machine learning (ML) can provide solutions to unlocking this data (for example by automatically identifying sensitive content or by creating metadata). However, AI also brings challenges in that it can compound and magnify pre-existing issues in collections (such as outdated and problematic language used in metadata) and reinforce bias – something that is a noted problem with AI in many applications across many sectors. In order to mitigate this, growing attention in AI research is being devoted to AI data preparedness, that is the systematic assessment, consolidation, documentation and screening of data meant to be used to train AI models.
The FLAME (AI For Libraries, Archives and Museums) project will address this particular issue in relation to the GLAM sector and will provide as the initial output from the project, open-access AI preparedness guidelines and an open-access article which will lead to significant impact:
· It will change the ways GLAM professionals prepare their digital records ahead of the application of AI;
· It will change the professional practices of digital archivists, with a focus on automation and the application of AI to archival collections;
· It will change the way general users and academics access records in archival collections and will move away from "dark" archives towards more accessible and environmentally sustainable collections.
Artificial intelligence can also enhance digital and digitised archives via the automatic description of image archives, automated transcription of archival texts and identifying names of people, places and organisations via named entity recognition (NER). This has already been demonstrated to greater or lesser degrees of success. However, most of the work being done in this area relates to Records Management eg: retention scheduling and appraisal, and although catalogue data has been mined by NER projects, there has been less interest in using the results to enhance cataloguing.
The AI and ML for Catalogue Conversion project aims to fill this gap.
Currently, most archives use the ISAD(G) system of cataloguing, however a new standard has been recently introduced – RiC (Records in Context) – which allows the relationship between entities to be characterised – for example an agent (e.g. a person) whose name might previously have been indexed (i.e. used as an entry point into the resource), can now, through RiC, be linked to a resource (e.g. a letter) by being its author, its recipient, its subject, its collector, its owner or its donor etc. Whilst NER can be used to find entities (in both ISAD(G) and RiC), it cannot be used to characterise their relationships. This project therefore aims to identify the potential for AI in identifying different types of relationships expressed within existing catalogue data or discoverable via AI from other resources.
This will clearly make a difference to those seeking to find information about a particular person or organisation and the aim of this project is to demonstrate proof of concept for the use of AI and ML to support the conversion of existing catalogue data from ISAD(G) to RiC via the incorporation of linked open data (LOD) resources.
This project, based in Liverpool, will also have a particular focus on identifying materials of potential relevance to the subject of enslavement. This is an area of huge current interest, both for academic and community researchers and a topic for which legacy finding aids are often inadequate. It is hoped that the outputs of the project would support Liverpool Record Office in identifying and providing better access to resources relevant to the involvement of the city in enslavement, supporting both academic and community researchers. The project would also support Liverpool University’s Centre for the Study of International Slavery, its partnership with National Museums’ Liverpool International Slavery Museum and their work with communities locally and internationally affected by the impacts of enslavement.
Importantly, it would also offer initial proof of concept for AI input to catalogue conversion. If successful, this would support the recordkeeping profession in moving from ISAD(G) to RiCs without significant additional staff resources. The ambition would be for any algorithms developed to be made open source or incorporated into existing software packages. It would also provide initial proof of concept for using AI to connect archival data to LOD. This has the potential to save work (individual repositories would not have to construct their own authority records) and would improve findability of resources. This would support enhanced collaboration between archival institutions holding related materials and cross-sectoral working between the GLAM professions.
Outputs from these projects will be made available in 2025 and will be disseminated via a range of channels including ARA’s own journal – Archives & Records, publications and bulletins.
Funding for these grants has been made available by the Archives and Records Association. ARA maintains reserves invested for the purpose of commissioning and supporting research, development and advocacy projects involving the creation, use, administration and conservation of archives and records and the means of developing the education of archivists, records managers and conservators, and articulating and advocating the unique identity, role and contribution of records and archives, as defined by its charitable objects. These designated funds constitute the Archives and Records Association Research, Development and Advocacy Fund. The Association’s Honorary Officers make grants from this Fund, on behalf of the Archives and Records Association Board.
You can read more about the initial call for proposals here: https://www.archives.org.uk/news/research-and-advocacy-grant-ai-and-emerging-technologies