Digital Humanities at Oxford Summer School 2026

Digital Safe: Creating a Trusted Repository for High-Security Content- Alex Kinnaman In addition to the treasures and data held by the Bodleian libraries that are in the process of being digitally preserved, there is an urgent need for unified, long-term preservation of University of Oxford records. Administrative, financial, medical, and personal records of University personnel are increasing rapidly on a daily basis. Digital Safe, created by the Electronic Archives Pilot Project, has sought to solve this problem by outsourcing the data management technology to Arkivum and managing it locally. This will allow all 38 colleges and other various departments to have a single technology for secure storage. Priority has recently been placed on completing Digital Safe, which calls for a final review of its trustworthiness, sustainability, and infrastructure. My project focused on conducting a comprehensive audit for Digital Safe. However, given that this data is not for openness and reuse and is in need of controlled, long-term storage, the metrics in current trusted digital repository audits are not entirely sufficient. Of the Assessment Tools approved by the Center for Research Libraries, TRAC was chosen to informally review Digital Safe for its universal use and thoroughness. While conducting a final review for Digital Safe has also revealed areas in TRAC that do not include aspects of non-traditional closed repositories. Further investigation has led this audit of Digital Safe to also act a first step to developing metrics to examine high-security, closed repository trustworthiness.

Redesigning and Extending the User Interface of the Semantic Alignment and Linking Tool (SALT)- Catherine Blauvelt Large scale dataset comparison is becoming increasingly relevant to current digital projects. Comparing these datasets often poses challenges, however, when identifiers, names, and language are not the same for each. The Semantic Alignment and Linking Tool (SALT) is a tool developed by David Weigl at the Oxford e-Research Centre that helps users establish connections between complementary datasets that lack common identifiers. SALT aligns these datasets by generating possible match candidates for confirmation or disputation by the user. The tool uses semantic context from the two datasets in question to prioritize candidates for alignment, as well as their extracted features, in datasets that would be difficult to compare otherwise. The extensive features and complexity of this tool require a straightforward and user-friendly interface. This project has focused on redesigning and extending the SALT user interface to incorporate derived features as additional semantic context. Drawing from user evaluations of the tool and from other similar tools, the new interface incorporates audio features provided by the FAST project and text derived features from the WCSA+DC project, and serves as a successful remodeling of a complex semantic alignment tool.