FAIR metadata and data in research projects

For who?

• Researchers who apply for and/or receive funding.
• Data stewards who assist with the grant application and/or the project or advise the researcher.

When?

• During the preparation of the project (during the grant application or during the writing of a DMP (after awarding the application).
• During the realisation of the project.

In which situations?

In the case of extra instructions from programmes (and calls for grant applications) for the production of FAIR metadata and data.

What does it cover?

1. General explanation about how ZonMw allows the production of FAIR metadata and data.
2. Specific instructions for the applicant or recipient of funding in the call for grant applications.

In the background information  it is stated that in a growing number of programmes, ZonMw encourages the application of FAIR principles during data management. Below we provide more information about FAIR metadata in connection with ZonMw's instructions for this in its calls for grant applications.

FAIR in brief

By applying FAIR principles, you make your data (or other sources for research) Findable, Accessible, Interoperable and Reusable. The outcome is that data are findable, and can be understood and used by people and computers. The approach to FAIRify your data is explained by Health-RI in The Metroline - Steps for your FAIRification Workflow, and by GO FAIR Foundation (GFF) in the ‘three-point-FAIRification’-framework’ (3-PFF).

Metadata schemes for ZonMw projects

Here we consider metadata in greater detail. Amongst other things, metadata describes where the data can be found, the context in which it was produced, the conditions for obtaining access to it and the information needed to be able to use it.

FAIR metadata means that the metadata can be read by a computer, in other words it is machine-readable’ or ‘machine-actionable’. Here we therefore talk about ‘metadata-for-machines’, abbreviated to M4M. Researchers and/or data stewards make metadata that describe the project and the data files (and any other possible sources for the research), that are used (‘used assets’) and generated (‘produced assets’) in it. Metadata are recorded in metadata schemes.

Domain-specific

Ideally, you use existing metadata schemes (metadata standards) that are tailored to a specific research topic, discipline or domain. This allows information about data from different projects to be compared.

The metadata schemas M4M Project Admin and M4M Dataset (catalogue, form and distribution) are generic, in other words the metadata elements (the ‘questions’) are the same for every project and every dataset. The metadata scheme M4M Project Content is domain-specific, because researchers and data stewards in the same research domain and/or consortium have made agreements together about the metadata elements with which they want to describe their datasets.

ZonMw has commissioned schemes for the subjects COVID-19, infectious diseases and antimicrobial resistance. These can be found on the  M4M - page of GO FAIR  Foundation. You can also find there schemes about subjects produced at the initiative of other organisations. The metadata schemes are open and can be used by everybody. If necessary, schemes can be expanded.

Learn more about the application of M4M metadata schemes in the COVID-19 program.

Under the hood

The metadata schemes we consider here have a few technical characteristics that are essential for making them machine-readable of machine-actionable:

• The metadata schemes and the answers of the user are readable for people in a questionnaire-like format. ‘Under the hood’ of the form, the information is in computer language and ‘persistent identifiers’ are linked to the information so that for the computer there can also be no misunderstanding about what people (the user of the metadata scheme) mean.
• Schemes work as much as possible with standardised control terms , the so-called controlled vocabularies. An extensive overview of these can be found on for instance BioPortal. For the answers to a question in the metadata scheme a list of terms appears that the user can choose from. That way the answers from all users can be properly compared and via a code (persistent identifier) are also understandable for the computer . An example of a standardised list (i.e. controlled vocabulary) is SNOMED to document and code medical data.

If you want to use a metadata scheme for FAIR data, make sure that the resulting metadata are ‘machine-actionable’.

Readable for both computers and people

The computer can find the metadata produced with these schemes, understand them and analyse them. To help people who are not at home in computer science to use the metadata,  they are exposed on a so-called metadata catalogue.

Likewise, the metadata forms for ZonMw’s COVID-19 programme can be found on the COVID-19 data portal of Health-RI.

Access to data

With most metadata catalogues it is possible to contact the data custodian, and send a request for permission to use the data. Some catalogues may even offer the service (under certain conditions) to directly obtain access to the data. With the help of metadata, the data custodian can precisely define who (or which algorithm) may or may not obtain access to the data. At the very least this is important for privacy-sensitive data. Metadata, in principle, contain no sensitive information and can therefore be made public.

Explanation for the grant applicant

In the cases that ZonMw requires grantees to produce machine-actionable metadata, this will be clearly indicated in the call for applications. If all other cases, this can be done on a voluntary basis. The activities for machine-actionable metadata are therefor in addition to the activities for regular data management. Grant applicants can prepare for this as follows:

• Determine  whether the data steward has experience with FAIR (meta)data, or has been trained for this.
• Determine whether metadata schemes for the discipline concerned already exist, and can be reused.
• Reserve  sufficient budget so that (1) the data steward and researcher can participate in activities (e.g. workshops) for the development of metadata schemes and (2) the data steward can help with filling these in. For this aspect and the 'standard' data management, about 3-5% of the project budget should be allowed.

Extra step toward FAIR data

ZonMw’s  FAIR metadata instructions initially concern the production of FAIR metadata that describe the context of the project and the dataset(s) used in the project. A next step in FAIRification could be for example to produce metadata for variables and units in the data file. Again, when this is required, ZonMw will indicate this in the call for grant applications.

Explanation for the project leader

If the project is awarded funding, then the project leader incorporates the planned activities and approach for FAIR (meta)data in the DMP, whether or not with extra instructions in the award letter. ZonMw informs the project leaders about any meetings that will be held. 

Approach and time estimation

In several ZonMw programmes, FAIR metadata schemes have been or will be developed that project leaders must use to describe their projects, datasets and/or other sources for research.  If ZonMw does not provide specific instructions for this, a project leader can also search for, adjust and/or develop machine-actionable metadata schemes on his own initiative.

More information about approaches and resources for making data and metadata FAIR can also be found in the FAIR Cookbook and RDMkit.