With the rise of Research Data Management (RDM), increasing amount of scientific data is organised and published, making scientific results transparent and enable data reusage. However, future reuse purposes are often not known, and data is still scattered in different places, especially in sciences with high degree of heterogeneity, like the engineering sciences. Data can be provided in data publication journals, in repositories (institutional/generic/discipline-specific), or even web pages. York Sure-Vetter, director of NFDI, states 2023 in an interview with the German Süddeutsche Zeitung: ''We are drowning in data, but cannot find it.' There is a lack of interconnected data spaces for science, he says, meaning protected virtual locations that facilitate the exchange of data across disciplines.' [translated by the author] The Data Mesh paradigm from industrial data management appears as a potential solution approach: Unlike other approaches, distributed data is not integrated into a central system, but interconnected from its original publication place. This let data remain in their original (specialised) repository instead of building a new complex monolithic system. As socio-technical approach, underlying organisational structures and requirements are reflected, going beyond a purely technical solution. Data Mesh is driven by four principles. First, Domain Ownership gives responsibility for the data to the creators instead of central IT teams, expecting to increase data quality by this. Second, product thinking is applied to data ('Data as a Product'), leading to a certain form of standardisation w.r.t. comprehensibility (e.g. by metadata schema, data model) and accessibility (e.g. by APIs) for future data reusage, even in context different than initially planned. Third, a central Self-Serve Platform is the entry point to maintain and find registered datasets, e.g. in a data catalogue. Fourth, within a Federated Governance global interoperability is ensured by common standards, while domain-specific rules can be implemented locally. The Data Mesh for RDM is proposed and designed here as research data publication infrastructure in the engineering sciences with its heterogeneity regarding data formats, data structures, and disciplines (like mechanical/electrical/civil engineering, etc.). It contributes to making data FAIR: - Datasets can be found in (one or multiple) data catalogues as well as in their repositories. - The Data Mesh redirects access to the original source, keeping authentication and authorisation where required. - Local and global governance specifies rules within domains and across the whole Data Mesh, fostering interoperability. - Finally, governance and the data product approach make data reusable. Moreover, a Data Mesh for RDM provides the potential to support interdisciplinary data-driven research in engineering and beyond. Existing data from various sources will be explored and provided within such a Data Mesh. Figure 2 depicts how existing components from RDM can contribute towards it. Architecture, concept, and use cases will be investigated in NFDI4ING's second funding period by the methodological-driven archetype Fiona. The idea already is described in detail in a journal article. This talk aims at spreading the vision in the RDM community beyond the engineering sciences and exchanging with other communities, in order to enable interdisciplinary research and identify components and activities that contribute to a Data Mesh for RDM.

The authors would like to thank the Federal Government and the Heads of Government of the Länder, as well as the Joint Science Conference (GWK), for their funding and support within the framework of the NFDI4ING consortium. Funded by the German Research Foundation (DFG) - project number 442146713.