Designing Data Infrastructure Maturity Profiles

This is a work-in-progress project. See the latest version athttps://shuai.ai/static/files/paper/DIMP.pdf. Data infrastructures (DI), in particular data repositories and portals, are critical in the ecosystem for research data to be Findable, Accessible, Interoperable, and Reusable (FAIR). Tools specifically designed for modeling DIs are lacking, complicating the comparison of implementation specifications like persistent identifiers, communication protocols, and authentication services across different infrastructures. Consequently, assessing an infrastructure's alignment with the FAIR principles and its adherence to other recommended guidelines is difficult without reviewing comprehensive documentation, which is often not accessible to the public. In addition, diverse features were found in DIs with different implementation priorities depending on the types of metadata/data and domain-specific requirements. For example, the PURE3D1 Research Infrastructure has a 3D interface of the object. However, the integration of metadata has not yet been completed. Some metadata of DANS SSH Data Station was collected by the ODISSEI Portal and Google Dataset Search but it remains unclear when, how often, and with what selection criteria3. Moreover, ODISSEI Portal's documentation misses details on how the collected metadata is refined, enriched, and republished. Less studied is software dependency, which encompasses the inherited characteristics arising from utilizing existing software components (e.g., PURE3D uses Voyager4) and platforms (e.g., Dataverse and CKAN5). An attempt towards a structured documentation of DIs is the re3data registry. By 26 April 2025, it includes 3,359 data repositories. For example, it provides essential features about the DANS SSH Data Station7 [9], including its repository size, repository type, policy, persistent identifiers, metadata standards, etc. However, it is not sufficiently detailed how it aligns with the TRUST principles for data repositories. For example, the need for transparency calls for more detailed documentation with an examination of dependencies of software components, tools, licenses, etc. Moreover, the link between implementation choices and the FAIRness of its data/metdata remains insufficiently articulated. A recent attempt using the FAIR Implementation profile (FIP) for the ODISSEI Portal demonstrated the potential of a more structured approach to documenting implementation specifications and how it can significantly facilitate analysis and research on DIs and their data/metadata. Documenting a DI's alignment with FAIR (software) principles is advantageous, as a DI is fundamentally software. This project addresses these challenges by providing a practical guide for assessing and advancing the maturity of data infrastructures. Inspired by existing FAIR assessment methodologies such as the FIP [14] and the FAIR Data Maturity Model, we introduce the Data Infrastructure Maturity Profile (DIMP), a structured assessment framework that maps high-level principles to concrete implementation decisions for the evaluation of DIs9. We systematically compare the above-mentioned principles/indicators/metrics and formulate implementation-driven questions. Furthermore, where necessary, we add questions regarding features captured by re3data, metrics proposed by previous DI assessments, checklists, and some software management plans (SMP). Finally, we adapt aligned questions to the context of DI design. By offering an actionable bridge between conceptual principles and operational practices, DIMP aims to be an instrument for documenting and comparing implementation choices, studying compliance with established principles, and benchmarking the progress of DIs and their implications on data management.