Version classiqueVersion mobile

White Paper — Open Science in a Digital Republic — Strategic Guide

Scientific and Technical Information Department - CNRS

Freedom to analyse scientific results

How to organise TDM?

Texte intégral

TDM structures and research centres in other countries

United Kingdom: The National Centre for Text Mining

1In the United Kingdom, the National Centre for Text Mining (NaCTeM) is a body financed by the state1 and overseen by the University of Manchester. It was set up in 2004, initially to address the needs of the university community.

2NaCTeM was set up to develop tools and services related to text mining.2 It offers text-mining services, software developed by NaCTeM teams or by others, seminars and workshops on TDM, tutorials and demonstrations, as well as publications resulting from text mining.

3Although this site was initially intended to assist the academic world in data mining, it now has a much wider scope and audience, including industry.3

4NaCTeM proposes to develop tailor-made tools and services to meet the needs of researchers in the academic or industrial worlds, when those available on the website are not suitable in the context of a particular project.4 An example of the services proposed by NaCTeM is FACTA+, which stands for Finding Associated Concepts with Text Analysis in the biomedical field:5

5NaCTeM strives to develop fully interoperable software tools, as the lack of interoperability is one of the greatest challenges facing researchers.6 The NaCTeM website does not list only the services and tools it has developed itself. TDM services and tools developed by third parties are also listed.

6NaCTeM also oversees projects7 relating to text and data mining and lists national and international events on the subject.8

7NaCTeM was the first national centre devoted to text and data mining.9

United States

8In the United States, the National Science Foundation (NSF) is an independent agency of the United States Government, whose mission is to fund basic scientific research. In 2002, $6 million were made available to support research in data mining in particular.10

9As for federal agencies, the Data Mining Reporting Act11 requires those who use or develop data mining to submit an annual report to Congress.12

10There seems to be no structure in the United States for sharing tools and data for purposes of text and data mining.

International Council for Science (ICSU) and World Data System (WDS)

11The International Council for Science (ICSU), created in 1931, is the largest non-governmental scientific organisation in the world. It is made up of 121 national members and 32 international scientific unions, and is responsible for encouraging international scientific and technological activities and advocating access to science for all, as well as preserving universal access to good quality scientific data for the long term.

12The ICSU created the World Data System in 200813 to promote and facilitate the exchange of data between the members of the WDS. The ICSU World Data System aims to make the transition from separate data centres to a global and interoperable data system, which would incorporate emerging technologies and the new data sciences. The mission of the ICSU WDS is to promote fair and universal access to reliable and good quality scientific information and data, in all disciplines.14

13In September 2016, the ICSU WDS had 98 members.15 In order to become a member, each candidate must be certified,16 in particular regarding the criteria of integrity and confidentiality:17

“As part of the process of developing WDS, a certification procedure for evaluating candidates for membership was developed by the Scientific Committee to ensure the trustworthiness of WDS Members in terms of authenticity, integrity, confidentiality and availability of data and services.”

14Data-Sharing Principles”, a sort of ethics charter, were drawn up, and include the principle of respect for the data source and its integrity:18

Data, metadata, products, and information should be fully and openly shared, subject to national or international jurisdictional laws and policies, including respecting appropriate extant restrictions, and in accordance with international standards of ethical research conduct.

Data, metadata, products, and information produced for research, education, and public-domain use will be made available with a minimum time delay and free of charge, or for no more than the cost of dissemination, which may be waived for lower-income user communities to support equity in access.

All who produce, share, and use data and metadata are stewards of those data, and have responsibility for ensuring that the authenticity, quality, and integrity of the data are preserved, and respect for the data source is maintained by ensuring privacy where appropriate, and encouraging appropriate citation of the dataset and original work and acknowledgement of the data repository.

Data should be labelled ‘sensitive’ or ‘restricted’ only with appropriate justification and following clearly defined protocols, and should in any event be made available for use on the least restrictive basis possible.”

The Research Data Alliance (RDA)

15The RDA is a joint initiative of the European Commission, the American National Science Foundation (NSF) and National Institute of Standards and Technology (NIST), as well as the Department of Industry, Innovation and Science of the Australian Government. Its purpose is to build a social and technical infrastructure to foster the openness and sharing of research data. The initiative is driven by the research communities themselves, and currently has more than 4,300 members from 111 countries. The members of the RDA are divided into Working Groups or dedicated Interest Groups, responsible for developing and validating an infrastructure and supporting the growth of a data-sharing community, including contributors from every scientific discipline and every geographical origin19.

16The RDA, launched in 2012, has already issued recommendations and tested and harmonised standards in different aspects of data science. These first results relate to aspects of infrastructure concerning the reproducibility of scientific output, long-term preservation and good practice in the maintenance of data repositories, the training of data specialists, citations of data, etc.

The framework proposed for European projects under H2020

17A political awareness of the need to provide a legal framework for TDM operations in the field of research has emerged:

  • in its strategy for a Digital Single Market, the European Commission had referred to the need to adapt the Directive on copyright (the InfoSoc Directive) to technological advances in order to avoid legal disparities within the single market;

    • 20 European Commission, Toward a modern, more European Copyright Framework, Brussels, 9.12.2015, COM(2 (...)

    a report of the European Commission from July 2016 stated that the absence of legal provisions at EU level concerning TDM for research purposes causes uncertainties with adverse effects on research and innovation (R&I).20

18These positions favourable to TDM have contributed to guiding Horizon 2020 towards open access and TDM, and the proposed Directive on Copyright in the Digital Single Market introduces an exception in favour of TDM.

19Horizon 2020 contributes indirectly to the development of TDM by fostering free access to research publications and data. This is because this European programme for research and innovation requires that all research publications and some data stemming from funded projects be released for free access after a certain period of time. Transparent and long-term management of research data must be ensured through the drafting of a Data Management Plan (for projects participating in the pilot scheme). To accompany this approach, the European Commission provides indications, published as guidelines, on good practice for managing research data.21 The European Commission’s strategy in favour of free access to research data is laid out in a report (Open Research Data Pilot22), published in December 2013. According to this document, Horizon 2020 projects must make every effort to enable third parties to access, mine, exploit, reproduce and disseminate the data from their research (free of charge for users).

20Horizon 2020 also contributes directly to the development of TDM by funding European projects on the subject. The European Commission invests in research on TDM with, for example, the H2020 call for projects GARRI-3-2014 “Scientific Information in the Digital Age: Text and Data Mining (TDM)”, issued in 2014. This call has given rise to several projects for the period 2015–2018, including:

  • FutureTDM,23 “The Future of Text and Data Mining”, which aims to spread the use of TDM in Europe. In particular, the project organises forums to obtain opinions and feedback concerning TDM from researchers, developers, publishers and other stakeholders.

  • OpenMinTeD,24 “Open Mining INfrastructure for TExt and Data”, which aims to facilitate the use of TDM technologies on scientific publications by making software and platforms interoperable via standardisation.

  • 25 All these data are taken from FutureTDM, Deliverable D4.1, European Landscape of TDM: Applications (...)

21The proportion of research projects on TDM among the projects funded by the European Commission has been calculated following a TDM analysis. Among more than 30,000 projects funded by the European Commission during the period 2007–2016 (FP7 and H2020), approximately 2.90% were related to TDM (n=885 projects).25 Some projects (0.38%, n=115) deal specifically with TDM. They all have the terms “Text Mining”, “Data Mining” or “Text and Data Mining” in their descriptions. The other projects (2.53%, n=770) contain similar terms, such as “Big Data”, “Data Analysis” or “Machine Learning”. This analysis also highlighted a semantic shift, with the expression “Text and Data Mining” being used increasingly.

22Horizon 2020 also proposes an ethical framework for research. Good research practice goes under the umbrella term of Responsible Research and Innovation (RRI). The goal is to take into account the potential impact of each innovation on society, anticipating future change and thinking in terms of sustainable development and sharing. In particular, personal data must be processed with full consideration of the right to privacy.

Data repositories in France: A few examples

The TDM framework in a pioneer discipline: Astronomy

23The abovementioned article “Préserver les données de la recherche à l’ère du Big Data” (Preserving research data in the age of Big Data) takes the example of the organisation of astronomy, a pioneer discipline in the conservation and sharing of data.

24The article quotes Françoise Genova, a researcher at the CNRS National Institute of Sciences of the Universe and the Strasbourg Astronomical Data Centre (CDS), who states that:

To understand the physical phenomena operating in astronomy, we need to collect observations obtained by different techniques and to work from data obtained by other instruments and other teams.”

  • 26 Guillaume Garvanèse, “Préserver les données de la recherche à l’ère du Big Data”, CNRS Le Journal, (...)

To respond to this need to exchange and preserve data, the international astronomical community is structured around the Virtual Observatory, a set of services that enables researchers to find useful information among all the astronomical data available to them through a large directory and shared standards.”26

25The Strasbourg Astronomical Data Centre (CDS) hosts the SIMBAD, the world astronomical reference database for the identification of astronomical objects. Its mission is to collect and exploit astronomical data and related information and make sure it is available all over the world. It is responsible for conserving data, maintaining the service and ensuring its long-term viability.

Huma-Num in the field of the human and social sciences

26“Huma-Num is a very large research infrastructure (VLRI) aiming to help research in the human and social sciences successfully navigate the move to digital.

27To fulfil this mission, the Huma-Num VLRI is built on a new type of organisation that involves mobilising both human and technological resources (by collective consultation and sustainable digital services, respectively) at the national and European levels, with the help of a large network of partners and operators.

28The Huma-Num VLRI aims to help consortia of players from within the scientific communities to coordinate the reasoned and collective production of source corpora (scientific recommendations, best technological practices). It is also developing a single technological structure for the processing, preservation, access and interoperability of research data. This set-up consists of a grid of dedicated services, a single access platform (ISIDORE) and a procedure for long-term archival.

29The Huma-Num VLRI also offers general good practice guides for researchers on the technology involved. It can provide expertise and training services on request. It participates in the DARIAH project on behalf of France as the National Coordinating Institution.

30The Huma-Num VLRI is run by the Joint Service Unit 3598, involving the CNRS, Aix-Marseille University and the Campus Condorcet.”27

The PREDON project

  • 28 Institut national de physique nucléaire et de physique des particules (French National Institute of (...)

31The PREDON project (for PREservation de DONnées, or Preservation of Data), an offshoot of the CNRS’s MASTODONS programme seeking to encourage cooperation between different disciplines around the concept of “Big Data”, was set up by a small group of researchers from the IN2P3.28 Its mission is to federate all French initiatives in the field of the preservation of scientific data. The project proposes a new approach based on the scientific, technical and organisational capacity of research units, international partnerships and major computing centres.

32The PREDON group works closely with similar French and international initiatives, especially with the International Committee for Future Accelerators (ICFA) panel for the preservation of data in high energy physics.

33In 2014, the PREDON Working Group produced a summary document (“Scientific Data Preservation 2014”) briefly describing the contributions of the participants in the workshops it has organised. The document is divided into three parts corresponding to the complementary aspects of the preservation of scientific data: scientific potential, methodology and technologies.

The expert contribution of the CINES

34The Centre Informatique National de l’Enseignement Supérieur or CINES (National Informatics Centre for Higher Education) is a national public body under the supervision of the Ministry of Higher Education and Research. The CINES has three strategic national missions:

  • supercomputing,

  • long-term archival of digital data,

  • hosting computing platforms on a national scale.

35As part of its second mission, the CINES is responsible for archiving the data and digital documents produced by the French higher education and research community, including scientific data from observation and experimentation, or resulting from calculation.

36It provides shareable, economic and customisable solutions for digital archiving in the medium and long term”,29 and “on the strength of its dual experience in supercomputing and long-term archiving, and also its excellent IT infrastructure”, the CINES “assists the producers and managers of data with their archiving problems”.30

37The main objective is to ensure the accessibility, integrity, readability and comprehensibility of the data for as long as necessary, by adapting the level of requirements necessary for the archiving as appropriate for the duration in question.

38In this framework, the CINES lobbies the players on the market to ensure that their file formats will remain readable for many years. A team of engineers is engaged in a permanent race against obsolescence by ensuring that software and hardware platforms can continuously access data. The FACILE web service31 lists the range of formats currently supported by the CINES.

A parallel with the way the hosting of healthcare data is organised

39To decide on the best way to organise the “preservation and communication” of files resulting from the processing of research data, lessons can be learned from the way the system for hosting healthcare data is organised in France.

40This is organised as follows:

41Creation. The ASIP Santé is a public interest group set up between the state, the CNAM (national health insurance fund) and the CNSA (national solidarity fund for autonomy).32 In 2009 the public interest group for personal medical records (GIP-DMP) became the ASIP Santé (Agency for Shared Information Systems for Health).33 The ASIP Santé is overseen by the Ministry of Health.34

42Role. The role of the ASIP Santé is to promote and supervise the development of shared systems in the areas of health and welfare sector.

43The need for interoperability. In order to facilitate and increase the exchanges between the different professionals in the areas of health and the welfare sector, interoperability must be achieved at the semantic, syntactic and technical levels.35 Making information entirely digital requires “the definition of languages common to all the information systems that will manipulate them, so that there is no need to develop new languages every time two information systems need to exchange or share data”.

44In compliance with its founding agreement, the ASIP Santé provides reference frameworks, standards, products or services contributing to the interoperability, security and use of health and telehealth information systems, and monitors their proper application. The Health Information Systems Interoperability Framework (HIS-IF) “sets the rules for a fully communicating health information system”.36

45Accreditation. A licence must be obtained for the hosting of health-related personal data, because of their sensitive nature.37 The ASIP Santé is responsible for the pre-appraisal of requests for accreditation and has developed a reference framework for completing the application forms.38 During this pre-appraisal, the ASIP Santé considers three aspects, detailed in the forms that the applicant has to complete:

  • an ethical and legal aspect (including the actions taken to obtain the consent of the persons concerned and the conditions relating to requests for corrections of information);

  • a security and technical aspect (including measures to guarantee security of access and data transmission, measures to control access rights and the traceability of access and processing, the conditions for the verification of hacking attempts and unauthorised access, and the means for verifying the registers of accredited persons and how they are kept up to date);

  • an economic and financial aspect.

46Accreditation is issued by the Ministry of Health for a duration of three years and is renewable.39

  • 42 Annual Report 2009 of the ASIP Santé, p. 29,

48Observations. The creation of a governance structure for the ASIP Santé has been welcomed by professionals in the sector of health and social information systems: “At last there is a clearly identified governing body working in a spirit of consultation.”42 Operators and actors within the sector are invited to participate in the development of the HIS-IF by expressing their needs, to ensure the relevancy of the repositories established by the ASIP Santé.

49The accreditation procedure for organisations wishing to host health data may seem onerous at first sight. Yet, on the one hand, it is necessary to ensure the security of the stored information in the light of its sensitive nature, and, on the other, the ASIP Santé attempts to make the process easier for applicants by making filing forms available. It is worth emphasising that this is simply a declaratory procedure, since only documents are examined. It should be noted that the Act for the Modernisation of the French Health System43 potentially provides for an assessment of technical compliance by a certifying body accredited by the COFRAC, instead of the approval procedure. Accordingly, on 16 September 2016 the ASIP Santé proposed a certification reference framework for consultation.44 Operators in the sector are thus invited to submit comments and proposals. It draws attention to the following points:45

  • the absence of any possibility of a posteriori control accorded by the Act to the General Inspectorate of Social Affairs;

  • the need for the reference framework to define the resources to be deployed and not only the goals to be achieved;

  • the consequent lack of visibility for applicants as to their chances of obtaining accreditation;

  • the need for human and technical resources for examining these applications;

  • the absence of outside audits by qualified auditors;

  • the need to take into account technical developments and the availability of commercial hosting services.

50An analysis of all the existing frameworks and actors and their mission and organisation enables us to sketch out and propose an organisational framework for the preservation and sharing of scientific research data for the particular purpose of TDM.












11 Section 804, Implementing the Recommendations of the 9/11 Commission Act of 2007, entitled The Federal Agency Data Mining Reporting Act of 2007 (Data Mining Reporting Act).









20 European Commission, Toward a modern, more European Copyright Framework, Brussels, 9.12.2015, COM(2015) 626 final.

21 European Commission, Guidelines on FAIR Data Management in Horizon 2020, 26 July 2016 (FAIR = findable, accessible, interoperable and reusable).




25 All these data are taken from FutureTDM, Deliverable D4.1, European Landscape of TDM: Applications Report, May 2016, p. 38.

26 Guillaume Garvanèse, “Préserver les données de la recherche à l’ère du Big Data”, CNRS Le Journal, 9 September 2016,


28 Institut national de physique nucléaire et de physique des particules (French National Institute of Nuclear and Particle Physics).





33 Decree of 8 September 2009 approving the founding agreement for a public interest group: ;




37 Article L. 1111-8 of the French Public Health Code:


39 Article R. 1111-15 of the French Public Health Code:;jsessionid=60C3A828598DFE6EA728751C24A32AF5.tpdjo15v_2?cidTexte=LEGITEXT000006072665&idArticle=LEGIARTI000006908152&dateTexte=20130118&categorieLien=id#vig



42 Annual Report 2009 of the ASIP Santé, p. 29,

43 Act No. 2016-41 of 26 January 2016 for the modernisation of the French health system (1), Article 204, 5th c: “I. - Under the conditions laid down in Article 38 of the Constitution and no later than two years after the promulgation of this Act, the Government is authorised to take the measures by ministerial order for the improvement and simplification of the health system falling within the scope of the Act, to: … c) Replace the accreditation provided for in the same Article L. 1111-8 by an assessment of technical compliance carried out by a certifying body accredited by the national accreditation body referred to in Article 137 of Act No. 2008-776 of 4 August 2008 on the modernisation of the economy, or by the competent body of another Member State of the European Union. This certification of compliance focuses in particular on the control of procedures, the organisation and the material and human resources as well as on the processes for the qualification of the hosted applications.”




Open access


Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search