Definition and principles

The Organisation for Economic Co-operation and Development (OECD) defines research data as "factual records (numerical scores, textual records, images and sounds) used as primary sources for scientific research, and that are commonly accepted in the scientific community as necessary to validate research findings".

The FAIR principles

Good data management aims to make data Findable, Accessible, understandable by humans and machines, i.e. Interoperable, and Reusable. This is what is called the FAIR principles. These principles cover the different ways that research data are generated, preserved, presented, shared, and reused.

• Assignment of a unique persistent identifier to the various data sets;

• Deposit in a data repository that is adapted to one's requirements (it is preferable if it is certified, known as a "trusted" repository, with guaranteed long-term preservation, the provision of a persistent identifier, and a system for managing data versioning);

• Creation of rich external metadata that is linked to datasets;

• Evaluation of the need for long-term archiving (with all of the steps that this entails).

• Data must be accessible to all authorised people, but no one else;

• Use standard, secure, free and open protocols when setting up databases;

• Always work, if possible, with file formats that are independent of any proprietary software;

• Metadata are always made available, even when the data are restricted or lost.

• To describe data, the metadata use controlled vocabulary that follow the FAIR principles (preferably, the documentation of the FAIR vocabulary used is findable and identifiable via its own persistent identifier);

• For optimal interoperability, choose a repository with linked data and constructed with RDF or equivalent technology, but this is not yet the most widespread option.

• A licence must be chosen to clearly stipulate the conditions for reusing the data. ;

• The data provenance must be described in detail;

• A "readme" file can be added if necessary to ensure proper understanding and future reuse of the data (ideally, it should also contain all the information needed to retrieve the data again under exactly the same conditions);

• The metadata vocabulary chosen must be appropriate for the types of resources, discipline, community, or repository.

The FAIR principles are presented in detail on the website of GO FAIR.

The issues involved in good data management

Managing data well is beneficial for you... and for others!

Good data management has three major advantages:

You save time: Getting started with documentation and implementing best practices take some time to get used to at the outset, but when integrated gradually throughout the project, they will save you time compared to complying at the end of the process, when the data is already far removed from its production context.
You limit the risks: Evaluating the resources used and considering the tools and methods for sharing enable you to produce and disseminate (with or without restricted access) in complete security. Your data remain identifiable and well-organised, and you avoid losses.
You make sharing and disseminating easier: Your data are stored in trusted, interoperable structures, making them easily accessible to both humans and machines. This gives your work greater visibility and impact.