Planned Preliminary Work with ‘ArchAI’ (Summer 2025)

Before beginning to work on ‘ArchAI’, I propose to interview scholars already using and designing digital resources for archaeologists in preparation for a scholarly article. As part of preliminary work, I will have several archival records containing descriptions of finds, drawings of missing site stratigraphy, and fragmentary site plans in the site archives across Rome, Pompeii, and Ostia. Significant work has been undertaken towards this goal over the past four years of my D.Phil. In preparation for the start of the fellowship and based on a thorough review of existing computational tools and approaches, I will then use the preliminary dataset and collaborate with colleagues at CUDAN and CNR to repurpose a custom data visualisation tool designed during my time at IU. The tool allows researchers to easily navigate local datasets for archival excavation data in a simple graphical user interface based on the framework of the UChicago and online annotation tools such as Hypothes.is, incorporates automated citation tools such as EndNote, and allows users to generate graphs quickly. The tool will aim to facilitate interdisciplinary collaboration and integrate with both the ‘Collection Space Navigator’ designed by CUDAN (Ohm et al., 2023), the ‘Extended 3D Matrix Tool’ developed by CNR (Lombardi et al., 2025), and datasets such as Manuscripts Online (Da Rold et al., 2024), Arachne (DAI, 2024), the Archaeology Data Service, HathiTrust, and the Heidelberg Archives. For my experiments, ‘ArchAI’ will leverage annotation and metadata structures for the Roman house designed during my dual-PhD and in collaboration with colleagues at IU, Oxford, Cambridge, UTexas, and the University of Michigan.

The completed version of the eventual desktop tool will allow scholars to integrate their own local dataset (often access-restricted) with quick and easy search tools for the annotation of local digitised archival excavation manuscripts and offer recommendations for related texts via the APIs of local archives (AAR, BSR, BH), superintendence and museum records (MDC, MANN, ICCROM, DigiVatLib), and existing collaborative databases focussed on archaeology (National Archives, HEIR, PompeiiInPictures, Ostia-Antica.org). Export options include creating a simple website that can be scanned and accessed via a QR code and edited online. The desktop tool (nicknamed ‘ArchAI’) will have a standard, visual, and easy-to-understand desktop interface, integrate with existing citation tools and be able to search both a local and private database of images, which are often access-restricted. Using a localised and expert-curated dataset will allow data to be modelled transparently and suggest possible connections for ‘experts in the loop’ to then evaluate (McClinton, 2020, pre-print). Relevant data will be manually curated and extracted by field specialists, and missing elements from the trained dataset will be automatically reconstructed, such as geometry in site plans or fragmentary texts. Source data will be clearly shown to the researcher and automatically included in exported visualisation and citations (Sommerschield et al., 2022). The resulting dataset will then be leveraged to visualise wide-scale changes to domestic spaces across Pompeii, Ostia, and Rome, prompting new research questions (Ahn, 2014). The resulting tool will be thoroughly documented and accessible as an open-access resource for other scholars to download on GitHub. Ongoing support and sustainability issues will be navigated with colleagues at Oxford, the Bodleian Libraries, and IU. While ‘ArchAI’ will be designed for archival research in domestic space, it will be valuable to a broader audience and provide a valuable collaborative and visual space, simultaneously exploring questions surrounding the use of computational tools in interdisciplinary and collaborative research.