The 'Federal Open Science Cloud' (FedOSC) – project (Belspo n.d.a, Chambers et al. 2024) is a two-year initiative (2024 – 2026) involving the Belgian Science Policy Office BELSPO, Belnet (Belnet 2020, Belnet 2024), Belgium's mandated organisation for the European Open Science Cloud (EOSC), and Belgium’s Federal Scientific Institutions (FSIs), to support the development of Open Science (Belspo n.d.c) and FAIR research data management (RDM) practices across the Belgian federal scientific landscape. Within these institutions, scientific research and cultural heritage management are often intertwined. (Belspo n.d.b, Perkins 2021) This contribution focuses on the GLAM dimension (Galleries, Libraries, Archives, Museums), particularly in the Arts and Documentation domains, and uses FedOSC as a use case to explore how data managed in GLAM contexts can be made visible and usable within Open Science environments.

While universities are usually well represented in most Open Science frameworks, knowledge within FSIs, especially in GLAM contexts, is less systematically captured in these environments (Rees Koerner et al. 2025). Cultural heritage data, such as images, catalogues, and collection records, is often already well maintained within domain-specific systems and infrastructures. However, these systems are typically not designed with research data reuse or Open Science integration as a primary objective. As a result, data that is produced for research purposes, or that could be directly reused in research contexts, often remains difficult to discover or connect to broader research workflows (Padilla et al. 2019).

The project combines policy development, service development, and direct institutional engagement. A core team of Data Stewards, organised into domain clusters (Art, Documentation, Nature, and Space), works closely with institutions either through embedded roles or via Single Points of Contact (SPOCs), including data managers, IT staff, collection managers, and researchers. This structure enables interaction with domain professionals and creates channels through which institutional practices, needs, and constraints can be articulated and addressed.

To better understand existing practices and needs across institutions, FedOSC conducted an online survey distributed through the SPOC network. The survey addressed both institutional and individual perspectives, covering the state of RDM practices, awareness of standards and support, and day-to-day RDM challenges. It also included questions aimed at identifying relevant datasets for further investigation. The results highlighted recurring issues such as limited time and capacity, unclear responsibilities for RDM, and gaps in metadata practices, particularly in GLAM contexts.

A central challenge addressed by FedOSC is the relationship between GLAM data and cross-domain Open Science practices. Cultural heritage data differs significantly from more standardised scientific datasets: it is heterogeneous, historically accumulated, and context-dependent, often reflecting interpretive decisions and institutional histories (Dressen 2025, Posner 2015). Much of this data is managed within collection-based systems rather than research data infrastructures and is not necessarily intended to function as research data despite its valuable scholarly content. FedOSC works within these conditions, supporting institutions in strengthening data management in their own contexts, while acknowledging that not all cultural heritage data can or should be treated in the same way as academic research data.

This approach translates into a set of coordinated activities at different levels. At the cross-institutional level, FedOSC is developing a data inventory that provides a structured overview of data types, formats, and reuse potential across institutions. This work builds on existing systems to make research-relevant data more visible across institutional boundaries and to support connections between collection-based infrastructures and research-oriented use cases. At the institutional level, Data Stewards engage with relevant stakeholders to assess current RDM practices, identify gaps, and discuss possible improvements. At the individual level, the project focuses on raising awareness of RDM concepts and available support, particularly among professionals in museums and collections who have not traditionally been the primary audience of Open Science initiatives.

A key methodological component of this work is the use of “data interviews.” In these interviews, Data Stewards engage directly with curators, archivists, researchers, and other domain specialists to document how data is produced, managed, and reused in practice. This includes identifying data types, formats, storage solutions, and existing guidelines, as well as understanding day-to-day challenges. These interviews generate structured insights into institutional practices while simultaneously providing an opportunity to offer practical guidance. In this sense, they function both as a research tool and as a form of embedded support.

Training and community-building activities complement this work by creating spaces for exchange across institutions. FedOSC has developed a growing learning community, supported through mailing lists, online platforms, and events, currently reaching around 200 participants per month. Training sessions, primarily in the form of webinars, address general Open Science topics while incorporating GLAM-specific examples and dedicated sessions, such as introductions to GLAM data and the concept of “collections as data.” These activities are designed to facilitate dialogue and translation between different terminologies, practices, and expectations across domains.

Early results show strengthened cross-institutional connections, particularly within the GLAM community, and a growing shared understanding of data practices. At the same time, engagement remains uneven and depends strongly on institutional support. In many GLAM institutions, responsibility for RDM is not clearly defined, and related tasks are often carried out informally, without dedicated time or resources. This can slow down implementation and limit the impact of Open Science initiatives. In addition, reaching the appropriate audiences remains a challenge, as participation is often driven by individuals who are already familiar with data management concepts.

These observations highlight that integrating GLAM into Open Science is not only a technical or infrastructural issue, but also an organisational and conceptual one. Cultural heritage data often contains rich contextual and interpretive information that does not easily fit into standardised frameworks. Preserving this context is essential for meaningful reuse, but it complicates efforts toward interoperability. Approaches to Open Science in this domain therefore need to balance these tensions, enabling visibility and reuse without flattening the complexity that gives heritage data its value.