From RDM Software to Semantic Portals - Leveraging Dataverse as a Flexible Framework for Research Data Discovery

Dataverse has long been recognized as a robust research data management (RDM) software solution, supporting the deposition, publication, and preservation of datasets across scientific domains [1]. And while Dataverse has long featured a modular metadata model supporting discipline-specific configurations, recent developments have significantly expanded its flexibility by evolving into an API-first architecture [2]. Together, these capabilities now position Dataverse not only as a research data management platform, but also as a framework for building customized research data services, such as semantic search portals supporting domain-specific findability, accessibility, interoperability, and reusability (FAIR) of research data. The Health Study Hub [3], Germany's central portal for medical research data within the national NFDI initiative, exemplifies this evolution. Originally built during the COVID-19 pandemic under rapid, needs-driven development, it has since transitioned to a more sustainable infrastructure with Dataverse as the backend. The Health Study Hub retained its dedicated single-page application (SPA) user interface, but now benefits from Dataverse's standardized APIs and flexible metadata management. The service brings together externally harvested studies and newly registered ones into a unified, searchable environment to enhance overall discoverability. The FAIRagro Search Hub, in development by the NFDI consortium for agrosystems research, builds on the same user interface but serves an entirely different domain. Firstly, it harmonizes dataset metadata from various agrosystem research data infrastructures, using a custom metadata model configured through Dataverse's extensible metadata blocks. This model enables domain-specific search capabilities, such as querying crop or soil information, and positions the service as the central entry point for FAIRagro's agricultural research data. Planned geospatial features will further enhance its relevance for location-based dataset discovery. Secondly, In addition to improving dataset findability, the FAIRagro Search Hub includes an integrated inventory of infrastructures relevant to agrosystems. Searchable via metadata, the inventory enhances the visibility of these infrastructures and supports cross-domain reuse. Planned features include service quality indicators, availability metrics, and summaries of data volume and content to help users identify and prioritize resources suited to their needs. Both portals use a custom user interface optimized for the target audience, rather than the standard Dataverse user interface. While this customization requires significant additional work, it ensures that the UI meets the needs of our users, placing a greater focus on rich descriptive metadata and intuitive search interfaces. It allows creating lightweight "filter pills" for refining search results in real time, and includes a query builder that supports the construction of complex Boolean expressions. These features hide the complexity of the underlying metadata structures, empowering researchers to explore data semantically and effectively. Dataverse's metadata management system and API-first redesign have thus enabled not just one, but multiple production-ready portals serving different scientific communities. They demonstrate how a shared infrastructure can support diverse use cases instead of building domain-specific siloed solutions. Importantly, both projects contribute back to the Dataverse ecosystem through feature requests, bug fixes, and direct contributions, strengthening the platform as a community-driven initiative. This abstract presents Dataverse not only as a reliable RDM system but as a sustainable and extensible foundation for semantic research data portals. By supporting customizable metadata models, flexible search interfaces, and active community engagement, Dataverse empowers institutions to build FAIR discovery services tailored to their domain needs, advancing the goals of national and international research data infrastructures.