Methodology for FAIRness Investigation to feature Multi-Dimensional Examination of Research Data Infrastructures in Agronomy
In the current era of big data and artificial intelligence, the Findability, Accessibility, Interoperability, and Reusability (FAIR) of datasets are foundational requirements for scientific progress. However, translating high-level FAIR principles into universal interpretable indicators remains a significant challenge. This is particularly evident in the multidisciplinary agrosystem domain, where decentralized Research Data Infrastructures (RDIs) were implemented for very specific use cases or demands. This often leads to fragmented FAIR properties that hinders overarching data discovery and integration. This is exacerbated by incomplete cataloguing of domain specific repositories. For instance, a considerable number of specialized German agrosystem RDIs are entirely absent from global registries, like re3data. To bridge these gaps, we propose a machine readable property catalogue that systematically documents repository features, including their core FAIRness properties. Complementing existing automated assessment tools [1-4], this talk will present a collaborative methodology with actively including RDI operators to complement their infrastructure’s metadata beyond what is currently registered in re3data, presenting the final dataset in the form of an interactively explorable RDI inventory.
We will show this from within the framework of the German FAIRagro by an initiative to revise and enrich information for RDIs in an inventory of repositories, relevant for agronomy. The central aim is to document their unique characteristics, and feature an interactive online catalogue and outlink to data search services. This catalogue requires continuous maintenance of the repository list and its metadata regarding FAIR properties, which is featured by a direct verification by the responsible RDI operators. This inventory service directly complements the global re3data registry, the recommended platform for NFDI consortia to register RDIs, in a collaborative approach alongside the FAIRagro Data. The proposed framework investigates these repository characteristics through a structured, multi-phase workflow rooted in the deliberate genesis, expansion, and curation of the domain inventory. The process is rooted by the collection of relevant RDIs and co-applicants central to the initial consortium proposal’s scientific use cases and work programme [5], followed by a preliminary annotation of FAIR-related metadata attributes by the proposal writing team. During the project’s lifetime, this foundational index was systematically extended by the FAIRagro Data Steward Service Center (DSSC) through expanded RDI tracking and deeper metadata annotation [6]. To transition these records into a machine-actionable format for the production-ready FAIRagro RDI inventory service, the compiled list was matched against re3data registry metadata records to harvest and harmonize existing annotations [7]. Concurrently, the workflow actively supported operators of FAIRagro RDIs not yet registered in re3data to facilitate their ingestion into the global index.
Building upon this established and verified inventory baseline, the proposed framework refines these initial repository FAIRness properties through a structured methodological workflow consisting of Metric Definition, interviewing RDI managers and the Community Validation. First, the framework adapts the RDA FAIR Data Maturity Indicators [8] into 20 core repository FAIRness metrics. Rather than acting as a rigid grading system, each FAIR category is structured around a 5-tier maturity scale that highlights technical milestones. This layout maps an infrastructure’s development from foundational operational readiness (1/5) toward advanced levels of machine-actionable integration (5/5), offering a transparent framework for community benchmarking and support. The framework was then operationalized into a structured interview deployed to RDI managers that constitute the extended list of FAIRagro RDIs. To ensure practical accuracy, the preliminary characteristics undergo an iterative Editorial validation loop, involving expert peer-review by data stewards from FAIRagro’s DSSC, as a panel of domain experts to help identify relevant RDIs not included so far and reviewing community suggested repositories.
The results of this investigation reveal across RDIs a high level of digital maturity in certain areas, such as legal reusability, while highlighting critical gaps in technical interoperability and accessibility. Next, we saw substantial improvement of the documented RDI properties by interviewing RDI operators. Specifically, our framework introduces completely new, machine-actionable attributes to evaluate elements entirely absent from the global schema, such as metadata persistence policies (F3), specific authentication protocols, the use of formal semantic knowledge representations and ontologies, and dataset provenance tracking. This allows us to transition from basic repository directory tracking to deep, multi-dimensional FAIR maturity scoring. Furthermore, we identified 35 RDIs not listed so far in re3data, out of which 15 were supported to get registered. Finally, we implemented an API to access the FAIRness metadata for RDIs. This granular API ensures that technical nuances of the RDIs may be monitored both by users and RDI maintainers. This allows RDI providers to easily observe existing operational gaps and identify precisely where their services can be improved. Ultimately, this framework transitions FAIR assessment from subjective exercises to an open, self-assessment-driven service. By publishing citable, transparent FAIRness features combined with expert editorial board-evaluations, this methodology reduces the technical burden for the Life Science consortia, fosters a culture of accountability and drives the sustainable adoption of FAIR standards in the decentralized agrosystems research landscape. Currently, a technical proof of these concepts is publicly available ( https://fairagro.github.io/rdi-fairness-interviews ) to showcase the frameworks evaluation capabilities, with full production integration into the centralized FAIRagro’s Repository Search Hub (ref) planned for the near future.
