Advancing data findability and re-use by PIDs at lower granularity levels

Traditional citation practices typically assign Persistent Identifiers (PIDs) only at the dataset or study levels. However, this approach fails to support the detailed citation required when researchers work with specific elements within datasets, such as survey variables or fragments of audio/video data. Data citation practices also vary widely , and researchers often do not follow a standard such as the Data Citation Principles (Joint Declaration of Data Citation Principles). As a result, researchers often cite data, especially dataset elements, in a free-text form, leading to semantic ambiguity, which makes it difficult to precisely determine which data was actually used in research outcomes. While PIDs on the dataset level are widely adopted, the absence of PIDs at a more granular level results in inconsistent citations, hindering reproducibility, transparency, data reuse, and discoverability. PIDs available at lower granularity levels ensure that each element, not just the entire dataset, is uniquely and persistently identifiable. This approach solves the widespread issue that researchers need to embed in their text complex referencing explanations of the specific data elements used. Instead, they can cite a PID that links to all necessary contextual information. This allows researchers to promote clarity in publications and peer review processes. Precise citations improve the credibility of published results, as peers can uniquely reference, verify and scrutinize the exact data used. Moreover, fine-grained PIDs enable findability of individual dataset components. Whether it's a single variable, a video snippet, or a textual excerpt, researchers can precisely find and identify what have been used. PIDs on dataset elements also enable granular citation tracking. This allows data creators to be appropriately credited for their contributions, which is increasingly essential for academic recognition and funding evaluation. KonsortSWD developed a PID Registration Service, based on the ePIC API and the Handle system, to address the challenge of fine-grained data citation in the Social Sciences. The scalable PID service enables Research Data Centers (RDCs) to assign PIDs to individual dataset elements. The service supports bulk registration, offers metadata validation aligned with established standards (e.g., schema.org, DDI), and ensures interoperability across systems by verifying and validating metadata content and format. Key achievements include the release of a fully functional prototype developed at GESIS - Leibniz Institute for the Social Sciences. A set of use cases for assigning fine-grained PIDs has been collaboratively implemented at several KonsortSWD partner institutions, including the German Institute for Economic Research (DIW/SOEP), the German Centre for Higher Education Research and Science Studies (DZHW), and Qualiservice. Scripts and workflows have been successfully implemented to assign test PIDs to diverse datasets, including smaller samples out of the 500,000 survey variables from GESIS, 100,000 from DIW, and 32,000 from DZHW; QualiService is currently testing the system to assign variables for qualitative data. Implementing PIDs at the element level within datasets offers a scalable, discipline-agnostic solution with immediate applicability across all NFDI consortia—making this service a cornerstone for interoperable, FAIR-aligned infrastructure across the whole of NFDI. Our PID service addresses a significant gap in current data infrastructures: the challenge of precisely identifying and citing the specific dataset elements used in research within different domains. The fine-grained PID model ensures that every data point—regardless of its format, volume, or origin—can be accurately identified, accessed, reused, and credited in a FAIR and sustainable manner.