Navigating the Data Patchwork - Strategies for Integrating Metadata Catalogs, Data Publications, and Archives

In modern science, the integrity and utility of research outputs critically depend on both data and metadata. Data comprise the primary outputs of experiments, simulations, or analyses—structured or unstructured information directly reflecting observed phenomena. Metadata, by contrast, describe the origin, structure, semantics, and context of those data, enabling their interpretation, discovery, and reuse. Although often treated as secondary, metadata are essential for ensuring compliance with the FAIR principles (Findable, Accessible, Interoperable, and Reusable), which have become a cornerstone of modern data stewardship [1]. A failure to adequately define or manage metadata leads to data silos, poor reproducibility, and limited interoperability [2]. To address this, we propose a solution architecture based on the decoupling yet tight linkage of a data and a metadata repository, following the principle of ``separation of concerns'' [3]. In this model, raw and processed data are published in scalable data repositories along with bibliographic metadata, while rich, structured, and extensible domain-specific metadata maintained in a dedicated metadata repository. Linking mechanisms (e.g. JSON-LD, RDF) together with persistent identifiers (e.g. DOIs, Handles) and metadata schemata (e.g. DataCite), ensureing integrity and support machine-actionable workflows across the research lifecycle [4], [5], [6]. At Helmholtz-Zentrum Dresden-Rossendorf (HZDR), this modular architecture is implemented through the integration of two distinct but interconnected components: the institutional data repository RODARE [7] and the domain-oriented metadata catalogue SciCat [8] (see Figure 1). RODARE is based on the Invenio framework [9] developed by CERN and provides a robust platform for the long-term archiving and dissemination of digital research objects, including datasets, software, and publications. It assigns persistent identifiers (DOIs), supports access control policies, and ensures compliance with institutional policies [10], [11] and FAIR data management guidelines. Complementing this, SciCat serves as a scientific metadata catalogue tailored to the needs of experimental and computational research workflows. SciCat captures rich, schema-based metadata at the time of data acquisition or processing, enabling fine-grained search, filtering, and workflow integration. The linkage between RODARE and SciCat is established through shared persistent identifiers, allowing each platform to serve its core function: SciCat for active metadata curation and workflow integration, and RODARE for long-term storage and publication of the datasets. %This separation of roles supports scalability, enhances metadata quality, and preserves provenance, while enabling automated synchronization and discovery across both systems. To streamline metadata generation and reduce manual curation effort, HZDR has developed a set of tools for automated metadata ingestion into SciCat. One key component for the scientific metadata is the electronic labbook at HZDR: A customized Semantic MediaWiki [12] instance, used by research groups to document experimental configurations, instrument settings, and contextual details in a structured, human-readable format. Through template-based forms and semantic annotations, this MediaWiki setup enables the extraction of relevant metadata elements, which are then programmatically transformed and ingested into SciCat via REST API. These automated pipelines capture domain-specific data at the point of creation, such as beamline settings, sample parameters or simulation inputs, and store them according to a community-specific schema. The additional subsequent provision of access to directories in the HZDR network is provided via HELIPORT [13]. In summary, the approach can serve as an illustrative example of a Research Data Management infrastructure combining separate data and metadata services into an overall concept.