Research data and their metadata should be FAIR, which stands for Findable, Accessible, Interoperable and Reusable.

The fundamental principle of FAIR data is "as open as possible, as closed as necessary".

The first requirement is that both people and machines can easily find the data. Therefore, the data:

• Receive a unique and persistent identifier, e.g. DOI.
• Organise the description of the dataset (metadata) into specific fields.
• The metadata is stored in some repository or database that is known or can be crawled by a search engine.

Once the data is located, the user needs to know how to access it. This includes authentication and/or information about access authorization:

• The software and protocols used for access are open and not proprietary.
• The metadata is preserved permanently in order to make it easier to contact the people responsible for it.
• Depending on the type of data, user registration and different access rights can be implemented.

Data are often integrated with other data, so they must be interoperable with other applications:

• Files and data must be represented in standardised, machine-readable computer languages and schemas.
• Datasets are described using controlled computer languages and unique, persistent identifiers.
• Datasets are linked to other datasets that complete them and the relationship between them is explained.

The ultimate goal of FAIR is to optimise the reusability of research data. To achieve this, it is essential that data and metadata are well defined. They must also be replicable and/or combinable with other datasets.

• The information about the data is detailed and relevant.
• Rights and use restrictions are made explicit.
• The description includes the origin of the data, those responsible for it, if they have been published, etc.
• Data and metadata are organised according to international standards for description, storage and exchange.