This presentation presents a practical approach to using artificial intelligence (AI) for tagging graduate theses housed in an institutional repository with the United Nations Sustainable Development Goals (UN SDGs). Utilizing strategies requiring no prior programming experience, this presentation will provide a step-by-step guide, cost analysis, and lessons learned from employing two AI-based tagging methods. These methods, attempted with varying degrees of success, highlight the potential of using AI for the thematic tagging of digital library resources.

Archipelago Commons, or simply Archipelago, is an open source repository platform developed and supported by the Digital Services Team at the Metropolitan New York Library Council (METRO). Archipelago features a curated list of open-source CMS and digital repository community built software and services, and the custom Archipelago Strawberry Field modules. All data, including metadata, for digital objects and collections is stored in JSON, and cast into different metadata schemas and displays using a unique templating system. In addition to standard metadata and display templates, IIIF API support and IIIF compliant viewer customizations are woven into Archipelago’s architecture and standard configurations. During this presentation, participants will learn how current IIIF Presentation API versions are implemented in an Archipelago repository through extensible templates and configurations. By reviewing our platform’s functionality areas related to IIIF integration, participants will learn about the simple, efficient way Archipelago repositories will be able to move forward to support the latest IIIF Presentation 4.0 API scheduled for initial release in the second half of 2025. Participants will also see a sneak peek at an example IIIF Presentation 4.0 manifest, and live demonstrations of customized implementations of popular IIIF viewers such as Mirador and IA Bookreader.

At CDL we work with librarians and archivists from across the ten University of California campuses. More specifically, as the team that maintains and augments our digital preservation repository, Merritt, we help depositors with their digital preservation projects. During recent collaborations, our team identified a need to be able to assist partner libraries with limited resources to stage their content ahead of ingest into the repository. This content is typically that which was generated as a result of work with a digitization vendor and resides on HDDs or other transfer devices.

To facilitate staging, validation and ingest of HDD content, we created a workspace that incorporates AWS S3, while simultaneously incorporating shell scripts and other resources to manipulate content once it is uploaded to the cloud. To this end, we implemented Amazon FSx for Lustre as a bridge between an EC2 instance and S3. The system also incorporates a Lambda function for submission automation purposes, as well as configuration management via Sceptre.

This workspace has allowed us to streamline the staging and ingest of newly digitized content while working with our library partners, via a collaborative solution that minimizes impact to resources of all teams involved.

Since 2017, 4Science has been working on implementing support for IIIF in DSpace to provide a better user experience in enjoying images, especially in the cultural heritage domain. To achieve this goal, a dedicated add-on has been implemented, easily integrated with a set of external Image Servers, such as Cantaloupe or Digilib. To enrich the content related to the digital cultural heritage managed within DSpace, we are now implementing workflows aimed at saving IIIF annotations created with Mirador and at relating them with all the information provided by metadata, fulltexts and entities. The proposed paper illustrates such workflows and how to relate annotations to each other and to other entities structured at data model level, in order to integrate them in the repository knowledge base.  

At OR 2024 I presented on work I had undertaken to add sustainable development goals (SDGs) to EPrints. The goal of this work was to a) present SDGs in an accessible and attractive way in EPrints repositories and b) relieve burden on repository administrators by incorporating automated identification of outputs related to SDGs.

This talk will discuss the progress towards this, more specifically an issue that arose of excessive and imprecise identifications by the automated searches that resulted from this search. Comparing the results of these searches in EPrints to those identified by the same search terms in Scopus, I will explore some of the possible reasons for these issues and some of the possible solutions. One specific area of investigation concerns the specific search engine and whether switching to a different mode of searching might alleviate this issue.

Creating FAIR (Findable, Accessible, Interoperable, Reusable) metadata is essential for research data reuse, yet it often poses a significant challenge for data depositors. At the Inter-university Consortium for Political and Social Research (ICPSR), located at the University of Michigan, we discovered a disconnect: while depositors are experts in their research, they struggle to translate this expertise into quality metadata. In response, ICPSR developed TurboCurator, an AI-driven tool that transforms depositor-provided content, like research summaries from publications/press releases and research plan methodologies, into metadata recommendations aligned with ICPSR’s standards. This presentation chronicles our journey to integrate responsible AI into the metadata creation process, blending machine intelligence with human expertise to enhance the depositor experience while upholding transparency and control.

The discoverability, attribution, and reusability of open research software are often hindered by its inadequate representation in research manuscripts. Frequently mentioned only implicitly or buried within supplementary materials, software fails to achieve recognition as a distinct, citable output. Addressing this challenge requires systematic identification and assignment of persistent identifiers (PIDs) to software, ensuring compliance with FAIR (Findable, Accessible, Interoperable, and Reusable) principles. Despite its significance, most open research software remains underrepresented in metadata, with limited explicit links between software and the research papers introducing or using them.

The SoFAIR project (2024–2025) seeks to enhance the identification and representation of software assets in research. By leveraging the global network of open repositories, the project aims to look into the current state of metadata standards and proposes adaptations to include software descriptions.

The presentation will explore current metadata formats and propose actionable solutions for improving the discoverability and reusability of open research software, aligning with best practices for metadata interoperability.