Persistent Identifiers and Knowledge Graphs - Investigating synergies between the NFDI basic services PID4NFDI and KGI4NFDI

CORDI25 RDM Infrastructure talk eventual DOI © Authors. This work is licensed under a Creative Commons Attribution 4.0 International License Persistent Identifiers and Knowledge Graphs Investigating synergies between the NFDI basic services PID4NFDI and KGI4NFDI Moritz Schubotz1, Sven Bingert2, and Tim O. F. Conrad3 1FIZ Karlsruhe – Leibniz-Institut f¨ ur Informationsinfrastruktur GmbH, Germany 2Gesellschaft f¨ ur wissenschaftliche Datenverarbeitung mbH G¨ ottingen, Germany 3Zuse Institute Berlin, Germany *Correspondence: Moritz Schubotz, [email protected] Abstract Persistent Identifiers (PIDs) and Knowledge Graphs (KGs) play crucial roles in the National Research Data Infrastructure (NFDI), advancing the FAIR principles for research data [1]. The PID4NFDI basic service builds sustainable infrastructures for the longterm identification of digital resources. KGI4NFDI supports building knowledge graphs that integrate datasets and enable semantic connections. Here, we briefly explore the synergistic potential between PID4NFDI and KGI4NFDI, aiming to identify shared strategies and future collaboration opportunities. A central theme of this synergy is linking PIDs to their respective entities within knowledge graphs, which enhances the discoverability of PID metadata and the data associated with these PIDs [2]. Identifiers are central to representing resources in knowledge graphs, as exemplified by Wikidata, where external identifiers enable the linkage of datasets across services and disciplines. One key challenge is to make the identifiers used in knowledge graphs (particularly those pointing to research data or FAIR Digital Objects) permanent. While PIDs are discoverable through systems like the PID-Meta Resolver (PID-MR), a challenge arises in identifying knowledge graphs listed in the KGI registry, a catalogue of disciplinespecific graphs in NFDI. Although KGI encompasses numerous smaller graphs that may include PIDs, there is currently no unified system capable of identifying all the relevant knowledge graphs that reference a given individual PID. This gap highlights the need for better interoperability between PID systems and knowledge graphs to foster cross-domain integration and improve data discovery and reuse. While centralized systems like the EOSC PID-Meta Resolver [3] have been used to resolve PIDs, the increasing number of discipline-specific PID initiatives [4], [5] raises questions about scalability. We draw on lessons from initiatives to assess the sustainability of these approaches and explore alternatives for efficient and automated PID registration [2] and automated registration for resolvers. Author et al. |ConfAbbrev Another important issue is defining what qualifies as a PID. We explore the criteria that distinguish PIDs from other types of identifiers and the challenges involved in ensuring that identifiers meet the rigorous standards set for PIDs such as those recommended by EOSC. We reflect on the experiences of services such as Wikidata, where the large volume of external identifiers poses challenges for determining which truly meet PID criteria. Finally, we investigate standardized protocols for metadata resolution. Existing PID resolvers typically offer limited redirection, restricting richer metadata discovery across services. We present an example from the Mathematical Research Data Initiative (MaRDI), which has developed a PID redirection system capable of providing more extensive metadata for each identifier. This approach demonstrates how combining PID systems with knowledge graph infrastructures — by linking PIDs to semantically enriched entity descriptions and interlinked metadata — can significantly improve the findability and interoperability of research data in accordance with FAIR principles. By combining the strengths of PID4NFDI and KGI4NFDI, these services offer a promising path forward for enhancing data interoperability, usability, and integration within the NFDI overall architecture [6]. This abstract highlights key technical challenges, opportunities for innovation, and potential directions for future collaboration between PID and knowledge graph infrastructures. Author contributions All authors contributed equally. Competing interests None. Funding This work was funded by FAIRCore4EOSC grant agreement id 101057264, MaRDI grant number 460135501, PID4NFDI, KGI4NFDI grant number 521466146. Acknowledgements We thank Torsten Kahlert for reviewing this abstract. References [1] M. Hofer, D. Obraczka, A. Saeedi, H. K¨ opcke, and E. Rahm, “Construction of knowledge graphs: Current state and challenges,” Information, vol. 15, no. 8, 2024. DOI:10.3390/ info15080509. [2] J. Nordling, P. Manghi, T. K´ alm´ an, et al.,F for findability: Persistent identifiers and knowledge graphs, Mar. 2025. DOI:10.5281/zenodo . 14988360. [Online]. Available: https: //doi.org/10.5281/zenodo.14988360. [3] S. Bingert and T. Zamani, Pidmr– eosc pid meta resolver, May 2024. DOI:10 . 5281 / zenodo.11192206. [Online]. Available: https://doi.org/10.5281/zenodo.11192206. [4] M. Bennett, M. Azzouz-Thuderoz, S. Malla Mohamad, DataCite, and F. K. .-. L. I. for Information Infrastructure, Zbmath <−>pidgraph integration demo dataset, Zenodo, Mar. Author et al. |ConfAbbrev 2025. DOI:10.5281/zenodo.15052569. [Online]. Available: https://doi.org/10.5281/ zenodo.15052569. [5] H. Cousijn, R. Braukmann, M. Fenner, et al., “Connected research: The potential of the pid graph,” Patterns, vol. 2, no. 1, p. 100 180, 2021, ISSN: 2666-3899. DOI:https://doi.org/ 10.1016/j.patter.2020.100180. [Online]. Available: https://www.sciencedirect.com/ science/article/pii/S2666389920302440. [6] M. Politze, P. Wieder, S. Schimmler, and M. Diepenbroek, Concept for setting up the overall architecture working group in the nfdi section ”common infrastructures”, Apr. 2025. DOI: 10.5281/zenodo.15210894. [Online]. Available: https://doi.org/10.5281/zenodo. 15210894.