Recommendation 1 – Require data to be cited
Research policy-making organisations and agencies should include a clear statement in their binding policy documents and recommendations requiring that research data be cited whenever they are used and mentioned. Publishing data (whether openly or under restricted access) and citing data should be considered essential for making research results replicable and the research process transparent. Organisations adhering to these policies should incorporate this requirement into their policies such as national open science policies, data policies, or PID policies. The policies should also suggest ways to monitor compliance and address instances of non-compliance.
Recommendation 2 – Include all six core components in the guiding policies
Through their policies, research policy-making organisations and agencies should stress that proper data citation – to fulfil the goals associated with citing data – requires at least the use of all the six core components. It should be noted that depositing data in a repository capable of providing the citation elements is a default expectation.
Recommendation 3 – Make the benefits of proper data citation clear in the policies
Scholarly citation requires that research data are referred to with the same rigour as other research outputs. In particular, policies that guide the openness, transparency, and reproducibility of research should emphasise that referring to research data not only promotes these goals but, in addition, makes it possible to increase the impact of science by referring to the data directly and by building various automated ways to enrich and create knowledge graphs when persistent identifiers are in place as required by the core components. Furthermore, policies should highlight that data citation practices are recognised as a key element of open and FAIR science and are increasingly incorporated into researcher evaluations and funding assessments.
Recommendation 4 - Require active reporting of data publication outputs through national or cross-national generic or field-specific infrastructures
Limited availability of metadata describing data publications creates obstacles for large-scale interlinking, monitoring and assessment of research outputs. When creating policies concerning research output reporting, policy makers should require RPOs to regularly report data publication metadata to relevant infrastructures such as research publication indexes, CRIS systems, grant tracking systems or library catalogues on a national or cross-national level. These infrastructures may be generic or cater to a specific research field.
Reported metadata should include at least the six core components of a data citation. If possible, it should also include information about additional contributors and organisations, related publications, research project identifiers, funder identifiers and grant IDs, using machine-actionable PIDs when feasible. Extensive availability of unambiguous and machine-readable metadata interlinking data publications with other entities will enable the implementation of scientific knowledge graphs as well as discovery and citation tools with widespread coverage.