Sharing research data more effectively

In compliance with its CNRS Roadmap for Open Science, the CNRS is encouraging its researchers to make their research data accessible1 in the same way as their scientific publications. Some scientists can access trusted thematic repositories2 or solutions their community have already developed. But what can they do if there are no such solutions available in their field? As a response to this question, the CNRS opened an institutional CNRS Research Data (French link) space on the Recherche Data Gouv platform.

"The CNRS is one of the major providers of research data on multidisciplinary themes. It is important to work on this issue as a whole - from big data to 'small' data that are very useful for science to progress", explains Sylvie Rousset, director of the CNRS's Open Research Data Department (DDOR).

The CNRS Research Data repository enables scientists to publish their data deriving from research supported by the CNRS. This repository offers a generic main collection with laboratories able to request the creation of specific collections. Around ten laboratories have already done so. Nearly 80 datasets are currently available in varied fields (chemistry, physics, engineering and systems sciences, sciences of the Universe, life sciences, etc.) representing over 6600 files.

• 1The 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".
• 2You can consult CAT OPIDoR to find the right solution for your data.

A comprehensive sovereign solution

This space was created as a state of the art solution by a consortium of institutions led by the National Research Institute for Agriculture, Food and Environment (INRAE)1 as a response to the requirements of the research community as a whole. "This space is part of the national Recherche Data Gouv ecosystem. This initiative has been highly structuring for French higher education and research, particularly the CNRS which is heavily involved in its development", explains Sylvie Rousset who is a member of the steering committee. "The early involvement of the CNRS means the whole community can benefit from the organisation's resources, skills and stakeholders that support the project. The CNRS deserves our gratitude for this", confirms Isabelle Blanc, ministerial administrator for data, algorithms and source codes at the Ministry of Higher Education and Research (MESR) which steers this ecosystem.

The Recherche Data Gouv space can be accessed through a single web portal and represents "a sovereign solution and a reliable alternative to commercial platforms for publishing data", according to Isabelle Blanc. In practical terms, this repository enables each institution to curate and moderate its data on its own institutional space like the one opened by the CNRS. Isabelle Blanc explains that the CNRS space "ensures that the maximum amount of research teams are not left without solutions as the CNRS is the supervisory authority for the most research units in France".

"France's ambition was set out in the 2016 Digital Republic Law – if 50% of a project's research work is publicly funded, There must at least be sharing and, at best, opening up of data2 ", she continues. A commitment to open up all data is also one of the criteria used by French and European funding agencies to evaluate projects and most publishers now stipulate that data linked to a publication should be accessible. "Recherche Data Gouv is part of this dynamic of open science, so that data can benefit the whole community", sums up Isabelle Blanc who also stresses the importance of the objective of "leaving no scientists without a solution for opening or sharing their data".

• 1The consortium includes the Inrae, CNRS and six universities (Grenoble Alpes, Lorraine, Strasbourg, Lille, Paris Cité, Paris Nanterre).
• 2Open data is data made available without control while data sharing can involve restricted access if the data is included in an exception set out in the 2016 law.

The priority is supporting scientists

"Opening up data is more complex than opening up publications. Research teams have to carry out a whole range of additional scientific work as far in advance as possible when designing their project. This can't be entrusted to third parties", explains Isabelle Blanc. Scientists need to be able to describe the tools, conditions and protocols used to produce and collect their data. An MESR survey run between 2018 and 2020 found that 80% of research communities either lacked support or the right infrastructure. The other countries1 the Ministry exchanged with on the subject had also identified a need for support in developing their national systems.

Support has become a central element of the system and a priority ahead of the development of technical solutions. In practice, support is organised through a network of services designed to respond to all requirements and based on a strategy of federating, promoting, enhancing and increasing existing initiatives. Firstly, the data management clusters are local solutions throughout the country which bring together complementary expertise from different establishments. Currently, 19 data management clusters are staffed by 350 people from 80 different institutions and provide a range of over 140 services. Ten more such clusters are planned and two calls for accreditation will take place in 2024 and 2025. "The CNRS is a partner in many data workshops which act as the main entry point for scientists", explains Sylvie Rousset.

• 1The United Kingdom, Canada, Australia, Norway and the Netherlands.

The CNRS also contributes to rolling out national resource centres like OPIDoR (French link) - a portal set up by the Inist-CNRS1 , a pioneer in the development of data management plans - and DoRANum (French link) which provides data sharing and management resources and training to support the scientific community. These two resource centres mean the CNRS has "a very special place" in the ecosystem, in Isabelle Blanc's view.

Most of the ecosystem's six thematic reference centres2 are also linked to the CNRS which is greatly involved in research infrastructures. This is the case, for example, of Huma-Num^IR* (French link) and Data Terra (French link). The latter works on the evolution of the Earth system, covering subject areas ranging from astronomy to climate change including the oceans, the poles, water resources and so forth. These subjects require the analysis of complex, highly heterogeneous observational data from multiple producers. This level of heterogeneity means such analysis often cannot be reproduced under exactly the same conditions. "The past situation was that good data management was an added value but now poor data management is a real obstacle to research", notes Nicolas Arnaud, director of CNRS Earth & Space. FAIR3 data is "part of the Institute's DNA" but such data now needs to be in the right place, at the right time and in the right format for all users working on these themes. For this reason, Data Terra is positioned downstream of data-producing infrastructures and provides a single portal for accessing data and dedicated data processing tools. In addition to the complex curation work carried out by this infrastructure, it also advises on best practices. "Data Terra is an object that functions and can be a source of inspiration for other communities, particularly on the European scale", explains Nicolas Arnaud.

• 1The Institute for Scientific and Technical Information (CNRS).
• 2Five thematic reference centres are directly attached to the CNRS: the Strasbourg Astronomical Data Centre, Data Terra, Progedo, Huma-NumIR* and the French Institute of Bioinformatics. The sixth is the French Biodiversity Data Hub of which the CNRS is a partner and which should be integrating Data Terra in 2024.
• 3Findable, accessible, interoperable and reusable.

Structuring the research data environment bolsters scientific collaboration and promotes interdisciplinarity through the sharing and re-use of data. Long-term human and financial resources are required to support the development of professions and skills. This is why Sylvie Rousset calls for "recognition of all these professions and profiles in career assessments". Currently, the skills associated with the FAIRisation of data are being effectively defined and efforts are being made to assign greater recognition to the specific nature of professions involved in managing, preserving and disseminating data like data librarians, curators or stewards, etc.

By 2025, the HAL open archive (French link) developed by the CNRS and the CCSD should also provide a service for direct deposits of datasets associated with publications that will then be made accessible via Recherche Data Gouv. Also that year, the Recherche Data Gouv ecosystem will apply to become a research infrastructure that is part of the national strategy (French link).

Recherche Data Gouv is currently also working towards European recognition (see box) and needs to find a sustainable economic model and governance system. To achieve this, a unit with its own staff and resources and operating under several supervisory authorities will be set up in 2024. Its financial support from the MESR will continue until 2026. Work on the harmonisation of legal information is also ongoing to support data teams and workshops on these complex issues which often need tailor-made solutions. "We hope this ecosystem can expand so it can provide this support and make the work of all institutions and research teams easier through a kind of snowball effect that will speed up the process", concludes Isabelle Blanc.