Strategic Research and Innovation Agenda (SRIA) of the European Open Science Cloud (EOSC) (v1.1, 2022)

- 1 - Strategic Research and Innovation Agenda (SRIA) of the European Open Science Cloud (EOSC) Version 1.1 – 1 November 2022 EUROPEAN PARTNERSHIP - 2 - Please note This SRIA version 1.1 was created to incorporate the MAR 2023-2024 into SRIA v1.0 and, hence, is not a full update of the EOSC Strategic Research and Innovation Agenda. During 2023, SRIA Version 2.0 will be developed, in concertation with the EOSC Community and the European Commission. Acknowledgements The original version of this document (SRIA v1.0) was written by the EOSC Executive Board [EOSC_EB] with the assistance of the EOSCsecretariat.eu project [EOSC_Sec], which supports the Executive Board in its activities. EOSCsecretariat.eu has received funding from the European Union’s Horizon Programme call H2020-INFRAEOSC-05-2018-2019, Grant Agreement No. 831644. The document developed, and incorporated feedback on the open consultation document, which was for the most part based on the EOSC Partnership Proposal [EOSC_PP] and drafts for the Strategic Research and Innovation Agenda for EOSC. Thanks are due to the members of the EOSC Executive Board Working Groups [EOSC_WGs] and all other EOSC stakeholders who have provided input to these Working Groups for their contributions to this document. This version, SRIA v1.1, was updated from v1.0 by the EOSC Association Board, who wishes to thank its Task Forces for their immense contributions under a tight timeframe to ensure the initial draft of the Multi Annual Roadmap (MAR) 2023-2024 was prepared in time for community consultation. The Board also wishes to thank all those who took time to read the document and respond to the consultation survey. Many constructive comments were received and have enabled us to produce a much more robust roadmap for Stage 2 (the period 2023-2024). Finally, the Board wishes to thank the EOSC Association Secretariat for proof reading and ensuring the text was placed into the appropriate template. Disclaimer The European Commission is not liable for any consequences stemming from the reuse of this publication. The views expressed in this publication are the sole responsibility of the authors and do not necessarily reflect the views of the European Commission. - 3 - Table of Contents FOREWORD BY DIRECTOR-GENERAL JEAN-ERIC PAQUET, DIRECTORATE-GENERAL FOR ‘RESEARCH AND INNOVATION’ ......................................................................................... 8 PREFACE BY KAREL LUYBEN, EOSC ASSOCIATION PRESIDENT (2020-2022) .................... 9 EXECUTIVE SUMMARY .................................................................................................... 11 HOW TO READ THIS DOCUMENT ..................................................................................... 17 1 NEW WAYS OF SCIENCE ........................................................................................... 19 1.1. The opportunity for change ........................................................................................... 19 1.1.1. Research in the digital age ...................................................................................... 19 1.1.2. European leadership ................................................................................................ 20 1.2. The request for change ................................................................................................. 20 1.2.1. From Gutenberg to Berners-Lee ............................................................................. 20 1.2.2. Lindau Declaration .................................................................................................. 21 1.3. Open Science ................................................................................................................. 21 1.3.1. Brief historical context ............................................................................................ 22 1.3.2. Open Science facets: documents, data and software .......................................... 24 1.3.3. Open Science adoption: progress and resistance ................................................. 29 1.3.4. Limits to Open Science ........................................................................................... 31 1.4. Next-generation infrastructure ..................................................................................... 32 1.4.1. Learning lessons from the recent past .................................................................. 32 1.4.2. Networking: the next-generation internet (NGI) .................................................... 37 1.4.3. Hardware: the computing continuum .................................................................... 37 1.4.4. Software: visualise, analyse, predict ...................................................................... 38 1.4.5. Data: findable, accessible, interoperable and reusable ........................................ 38 1.4.6. Machines for scientists: EOSC foundations .......................................................... 39 2 EOSC IN THE MAKING .............................................................................................. 41 2.1. European Research Area ............................................................................................... 41 2.2. Priorities of the new Commission ................................................................................ 43 2.3. The European strategy for data .................................................................................... 43 2.3.1. Europe-wide common data spaces ........................................................................ 44 2.4. From Horizon 2020 to Horizon Europe ........................................................................ 46 2.5. A history of EOSC .......................................................................................................... 49 2.5.1. Transition period 2019-2020 .................................................................................. 50 2.5.2. National infrastructures across Europe ................................................................. 52 2.6. International dimension ................................................................................................ 54 2.7. Strengthening the community ...................................................................................... 55 2.7.1. EOSC Association and EOSC Partnership .............................................................. 55 2.7.2. Governance of the EOSC Association .................................................................... 58 3 STRATEGIC OBJECTIVES OF THE EUROPEAN OPEN SCIENCE CLOUD ..................... 61 3.1. EOSC Objectives Tree .................................................................................................... 61 3.2. Ensure that Open Science practices and skills are rewarded and taught, becoming the ‘new normal’ ............................................................................................................................... 62 - 4 - 3.3. Enable the definition of standards, and the development of tools and services, to allow researchers to find, access, reuse and combine results ......................................................... 63 3.4. Establish a sustainable and federated infrastructure enabling open sharing of scientific results ......................................................................................................................... 64 4 GUIDING PRINCIPLES .............................................................................................. 67 4.1. Introduction ................................................................................................................... 67 4.2. Multi-stakeholderism .................................................................................................... 68 4.3. Openness: ‘as open as possible, as closed as necessary’ .......................................... 70 4.3.1. Open access ............................................................................................................ 71 4.3.2. Trust in science through improved reproducibility ................................................ 71 4.3.3. Facing global challenges through multi-disciplinary programmes ...................... 73 4.4. FAIR guiding principles: making science transparent and reproducible .................... 73 4.4.1. Web of FAIR Data and Related Services for science ............................................. 73 4.4.2. Diversity of FAIR practices ...................................................................................... 74 4.4.3. Community standards ............................................................................................. 74 4.4.4. Research artefacts sustainability ........................................................................... 75 4.4.5. FAIR metrics and certification ................................................................................ 75 4.5. Federation of infrastructures ........................................................................................ 76 4.5.1. First iteration – Minimum Viable EOSC ................................................................. 76 4.5.2. EOSC-Core ................................................................................................................ 77 4.5.3. EOSC-Exchange ....................................................................................................... 78 4.5.4. Federated data and services .................................................................................. 79 4.5.5. Future Outlook ......................................................................................................... 81 4.6. Open Science services: machines in support of people .............................................. 84 4.6.1. Digital systems for Science .................................................................................... 85 4.6.2. Hardware .................................................................................................................. 85 4.6.3. Software ................................................................................................................... 85 4.7. Recommendations ........................................................................................................ 89 4.7.1. Fund awareness-raising, training, education and community-specific support .. 90 4.7.2. Fund development, adoption and maintenance of community standards, tools and infrastructure ................................................................................................................ 91 4.7.3. Incentivise development of community governance ............................................ 92 4.7.4. Translate FAIR guidelines for other digital objects ............................................... 93 4.7.5. Reward and recognise improvements of FAIR practice ....................................... 93 4.7.6. Develop and monitor adequate policies for FAIR data and research objects ..... 94 5 IMPLEMENTATION CHALLENGES ............................................................................ 96 5.1. Identifiers ....................................................................................................................... 96 5.1.1. Status ....................................................................................................................... 96 5.1.2. Gaps ......................................................................................................................... 97 5.1.3. Priorities ................................................................................................................... 98 5.2. Metadata and ontologies .............................................................................................. 99 5.2.1. Status ....................................................................................................................... 99 5.2.2. Gaps ...................................................................................................................... 100 5.2.3. Priorities ................................................................................................................ 100 5.3. FAIR metrics and certification ................................................................................... 101 5.3.1. Status .................................................................................................................... 101 - 5 - 5.3.2. Gaps ...................................................................................................................... 102 5.3.3. Priorities ................................................................................................................ 103 5.4. Authentication and authorisation infrastructure ...................................................... 104 5.4.1. Status .................................................................................................................... 104 5.4.2. Gaps ...................................................................................................................... 105 5.4.3. Priorities ................................................................................................................ 107 5.5. User environments ..................................................................................................... 107 5.5.1. Status .................................................................................................................... 107 5.5.2. Gaps ...................................................................................................................... 109 5.5.3. Priorities ................................................................................................................ 110 5.6. Resource provider environments ............................................................................... 110 5.6.1. Status .................................................................................................................... 110 5.6.2. Gaps ...................................................................................................................... 111 5.6.3. Priorities ................................................................................................................ 112 5.7. EOSC Interoperability Framework .............................................................................. 112 5.7.1. Status .................................................................................................................... 112 5.7.2. Gaps ...................................................................................................................... 113 5.7.3. Priorities ................................................................................................................ 116 6 BOUNDARY CONDITIONS ....................................................................................... 119 6.1. Rules of Participation ................................................................................................. 119 6.1.1. Status .................................................................................................................... 119 6.1.2. Gaps ...................................................................................................................... 119 6.1.3. Priorities ................................................................................................................ 120 6.1.4. Considerations ...................................................................................................... 121 6.2. Landscape monitoring ............................................................................................... 123 6.2.1. Status .................................................................................................................... 123 6.2.2. Gaps ...................................................................................................................... 124 6.2.3. Priorities ................................................................................................................ 125 6.3. Funding models .......................................................................................................... 126 6.3.1. Status .................................................................................................................... 126 6.3.2. Gaps ...................................................................................................................... 127 6.3.3. Priorities ................................................................................................................ 128 6.4. Skills and training ....................................................................................................... 128 6.4.1. Status .................................................................................................................... 128 6.4.2. Gaps ...................................................................................................................... 130 6.4.3. Priorities ................................................................................................................ 130 6.5. Rewards and recognition ........................................................................................... 134 6.5.1. Status .................................................................................................................... 134 6.5.2. Gaps ...................................................................................................................... 134 6.5.3. Priorities ................................................................................................................ 135 6.6. Communication .......................................................................................................... 137 6.6.1. Status .................................................................................................................... 137 6.6.2. Gaps ...................................................................................................................... 137 6.6.3. Priorities ................................................................................................................ 138 6.6.4. Considerations ...................................................................................................... 138 6.7. Widening to public and private sectors and going global ........................................ 139 6.7.1. Widening to public and private sectors ............................................................... 139 - 6 - 6.7.2. Going global .......................................................................................................... 144 7 EXPECTED IMPACTS ............................................................................................. 146 7.1. Improved trust, quality and productivity in science .................................................. 146 7.2. Development of innovative services and products ................................................... 149 7.3. Improved impact of research in addressing societal challenges ............................ 151 7.1. Critical success factors ............................................................................................. 153 8 ROADMAP .............................................................................................................. 154 8.1. Results of the EOSC open consultation .................................................................... 155 8.2. Minimum Viable EOSC (MVE) .................................................................................... 157 8.2.1. Components of the MVE ...................................................................................... 157 8.2.2. Scope and timing of the MVE .............................................................................. 160 8.2.3. H2020 and Horizon Europe EOSC projects ......................................................... 161 8.3. Objectives ................................................................................................................... 162 8.4. Levels of implementation .......................................................................................... 166 8.5. Priorities Stage 1 (2021–2022) ................................................................................. 167 8.5.1. Objective 1 ............................................................................................................ 167 8.5.2. Objective 2 ............................................................................................................ 168 8.5.3. Objective 3 ............................................................................................................ 169 8.6. Priorities Stage 2 (2023-2024) ................................................................................... 170 8.6.1. Objective 1 – Open Science as the new normal ................................................. 172 8.6.2. Objective 2 – Definition and development of standards and tools ................... 175 8.6.3. Objective 3 – Establish a sustainable and federated infrastructure ................. 179 8.7. Building on the Horizon Europe 2021-2022 calls ...................................................... 182 8.7.1. Open science as the new normal ........................................................................ 182 8.7.2. Definition and development of standards and tools .......................................... 183 8.7.3. Establish a sustainable and federated infrastructure ........................................ 183 8.8. Key performance indicators ....................................................................................... 184 9 CONCLUSIONS ....................................................................................................... 186 APPENDIX A RELATED DOCUMENTS ....................................................................... 191 REFERENCES ................................................................................................................. 202 LIST OF ABBREVIATIONS .............................................................................................. 213 GLOSSARY .................................................................................................................... 217 - 7 - Table of Figures Figure 0.1: European Open Science Cloud Objectives Tree ...................................................... 12 Figure 1.1: Research activity flows ............................................................................................. 22 Figure 1.2: Open Science taxonomy (from the FOSTER project) .............................................. 23 Figure 1.3: Research lifecycle and Open Science (from the FOSTER project) ......................... 24 Figure 1.4: Open Science facets (from the FOSTER project) .................................................... 25 Figure 1.5: Open Science at the crossroads between communities and funders, organisations, and ministries ............................................................................................................................... 30 Figure 1.6: The Computing Continuum (from the ETP4HPC SRA) ........................................... 37 Figure 3.1: European Open Science Cloud Objectives Tree ...................................................... 61 Figure 4.1: Schematic representation of key elements of the Minimum Viable EOSC ............ 77 Figure 4.2: Schematic representation of timelines of EOSC iterations .................................... 83 Figure 4.3: Software ontologies landscape derived from Pathways for Discovery of Free Software [Gruenpeter & Thornton] CC-by-4 ................................................................................ 87 Figure 4.4: Cloud Computing layers (from Wikipedia) ............................................................... 88 Figure 4.5: Cloud Computing types ............................................................................................. 88 Figure 5.1: Composite image of searches for IVOA resources in B2FIND ............................ 115 Figure 6.1: Actors in the EOSC ecosystem: roles and interactions ........................................ 129 Figure 6.2: Roles of data stewards in the data stewardship landscape in Denmark and the Netherlands [DS_Roles] ............................................................................................................ 132 Figure 7.1: EOSC Objectives Tree – benefits ........................................................................... 146 Figure 8.1: Analysis of open consultation on the SRIA ........................................................... 156 Figure 8.2: Scope and timing of the MVE ................................................................................ 157 Figure 8.3. The Minimum Viable EOSC .................................................................................... 158 Figure 8.4. Scope and timing of the MVE ................................................................................ 160 Figure 8.5: Duration of previous and existing EOSC projects ................................................. 162 Table of Tables Table 1.1: How to read this document ........................................................................................ 17 Table 1.1: EOSC in its technological context .............................................................................. 34 Table 4.1: Commonly used systems, of which EOSC offers an integrated view by federation ....................................................................................................................................................... 85 Table 4.2: Overview of recommendations and the stakeholder groups to which they apply . 90 Table 8.1: General Objectives (from SRIA v1.0 and interpretation) ....................................... 163 Table 8.2: Specific Objectives (from SRIA v1.0 and interpretation) ....................................... 165 Table 8.3: Operational Objectives (from SRIA v1.0 and interpretation) ................................. 166 Table 8.4: KPIs for expected outcomes, by strategic objective ............................................. 185 Table 9.1: List of documents related to the EOSC SRIA ......................................................... 191 - 8 - Foreword by Director-General Jean-Eric Paquet, DirectorateGeneral for ‘Research and Innovation’ 1 The Commission Communication on a new ERA for Research and Innovation [EC_COM_New_ERA] stresses the importance of Open Science as the most efficient and effective way of carrying out research to increase knowledge circulation, open collaborative work and the wide and rapid sharing of research outputs. The ultimate objective of Open Science is to increase scientific quality, the pace of discovery and technological development, as well as societal trust in science. The Communication highlights, specifically, the role of the European Open Science Cloud (EOSC) [EOSC] to drive a well-functioning and highperforming European R&I ecosystem by fostering the flow of research data and scientific knowledge between researchers, institutions and disciplines. The challenges that our society faces, such as the COVID-19 pandemic and the climate emergency, are increasingly complex, cross-border and inter-sectoral in nature. The research needed to address them is often data intensive, spanning many disciplines and countries. Long gone is the era of the lonely scientist. Now is the time for collaborative science, which is open, trusted and digitally enabled. The digital outputs of research (publications, data or software, among others) must be shared widely and as early as possible between researchers, but also between communities and with society. This requires that these outputs are made FAIR (findable, accessible, interoperable and reusable) [GO FAIR_Principles] and that the related data services are available. This is what EOSC is about: developing, deploying and evolving a trusted environment providing two million European researchers with seamless access to research data, research infrastructures, e-infrastructures and related services, enabling them to share, curate, discover, access, process and reuse research outputs of all kinds across borders and scientific disciplines. This Strategic Research and Innovation Agenda (SRIA) provides a clear roadmap over the next seven years to achieve the EOSC vision and objectives. It results from a collective, forwardlooking co-creation process to identify and prioritise complementary activities at EU, national, and institutional levels. Its content has been validated through wide consultation of EOSC stakeholders, including representatives of EU Member States and other countries associated to the EU R&I programme, research-performing and research-funding organisations, research infrastructures and e-infrastructures, research libraries and research associations. I want to warmly congratulate the members of the EOSC Executive Board and the many experts who worked hard to drive the SRIA development. Thanks to their remarkable professionalism and dedication, their efforts have been successful. I look forward to a fully operational EOSC that delivers the ‘Web of FAIR Data and Services for Science’. This will be essential to effect the digital transition in the European Research Area, enabling data-intensive interdisciplinary research and innovation, and accelerating the process of scientific discovery and technological development. 1 Mr. Paquet served as Director-General of DG RTD from 1 April 2018 until 1 September 2022. - 9 - Preface by Karel Luyben, EOSC Association President (20202022) This Strategic Research and Innovation Agenda (SRIA) is the latest in a series of increasingly significant steps towards making the European Open Science Cloud (EOSC) a reality. A collaborative effort, the development of the SRIA is an example of the multi-stakeholder commitment, cooperation and consensus that are essential to realising EOSC’s vision and potential. EOSC was conceived as the solution to the problem of how to manage and exploit the unprecedented volume of data arising from digital technologies, in the interest of European science and scientists. By providing an open and trusted environment for accessing and managing a wide range of publicly funded research data and related services and complementary commercial services, EOSC will transform how researchers access and share data throughout the research lifecycle, helping European scientists reap the full benefits of data-driven science and giving Europe a global lead in both research data management and scientific progress. It will federate existing dataand e-infrastructures, currently dispersed across disciplines and the EU Member States, around a federating core, with frameworks, principles and rules – many of them based on already established initiatives – to ensure the data are findable, accessible, interoperable and reusable (FAIR), while respecting data sovereignty, security and data protection requirements and regulations. Its role is to facilitate and optimise the sharing of existing data, digital objects, services and resources, not to reinvent; it is driven by the needs of science, scientists and the wider stakeholder community, not by technological advancement for its own sake. To date, multiple projects funded by Horizon 2020 calls have helped to lay foundations for EOSC, engaging a wide range of stakeholders and communities and delivering relevant outputs including use cases, demonstrations, data service tools, policy documents and, most visibly, the EOSC Portal. Progress towards enabling the creation of the EOSC ecosystem has thus been significant – yet fragmented. Beginning with the transition governance structure and culminating in the creation of the EOSC Association, steps have been taken to overcome that fragmentation in the future, to provide a framework for consolidation, coordination and collaboration, and to bring the community together at European, national, regional and institutional levels. As EOSC enters the next implementation phase under Horizon Europe, the commitment, cooperation and consensus of all stakeholders are key, from researchers through service providers and infrastructure operators to funders and policy makers. The EOSC Association – with, at the time of writing, 142 members (21 of them nationally mandated organisations) and 49 observers representing the full spectrum of stakeholders – is well placed to deliver that commitment, cooperation and consensus, while its Co-programmed European Partnership with the European Commission provides the optimum vehicle for realising the vision outlined in this SRIA. The challenges involved are immense, and are acknowledged throughout the SRIA, not least in the critical success factors and 14 action areas it enumerates. But the societal benefits - 16 - this version of the SRIA focuses on the first implementation stage, 2021–2022, which aims to develop added value from a federation of infrastructures by providing the core functions of the Minimum Viable EOSC (MVE) that will enable EOSC operations (the EOSC-Core). For each strategic objective a set of priority activities is defined, together with the most appropriate levels of implementation (European, national, institutional) and expected outcomes. Key performance indicators (KPIs) are also defined, with tentative target levels of accomplishment and dates, and measurement method (direct, survey). The work plan for Stage 1 will set the foundations for the subsequent roadmaps for Stage 2 (period 2023–2024) and Stage 3 period 2025–2027), and the work plan for Stage 2 will prepare us further for Stage 3. Ultimately, EOSC will deliver a research environment that promotes Open Science and increases trust and reproducibility in research outcomes. The overall impact is a pan-European research landscape that offers significantly improved discovery, access, interoperability, and exploitation of research outputs for researchers and research and innovation stakeholders. In conclusion, EOSC is well placed to deliver Europe’s deployment of Open Science within the Horizon Europe work programme and beyond. The legal entity, the EOSC Association, with 27 nationally mandated organisations, 133 members, and 78 observers, involves research and innovation stakeholders both across and outside of the EU. The Association provides a sustainable means, recognised by the EC, to serve and strengthen the EOSC community, coordinate the identification of needs for the development of EOSC, promote alignment of EOSC contributions at all levels and support Open Science development in Europe. - 17 - How to read this document This Strategic Research and Innovation Agenda (SRIA) contains a wide range of information – both historical and forward-looking, high level and detailed, aspirational and practical – and can therefore be read either in its entirety or in part, depending on the interests and/or needs of the reader. Suggestions for approaching the document are summarised in Table 1.1. The sections that relate most closely to future research, development and innovation activities in respect of the European Open Science Cloud (EOSC) are highlighted. If you are interested in . . . Then please see . . . A standalone summary of the document Executive summary The history of the digitisation of research – the science-supporting technological context from which EOSC has evolved Section 1 New ways of science The recent EC/EU/ERA policy context for Open Science and Open Data and the development of the EOSC initiative and its governing bodies Section 2 EOSC in the making The strategic objectives that are driving EOSC Section 3 Strategic objectives of the European Open Science Cloud The guiding principles and recommendations that are shaping EOSC Section 4 Guiding principles The primarily technical challenges and prerequisites to implementing the EOSC ecosystem Section 5 Implementation challenges The social, financial, legal, educational, cultural challenges and prerequisites to implementing the EOSC ecosystem Section 6 Boundary conditions The anticipated benefits of EOSC and critical success factors Section 7 Expected impacts The priorities for EOSC in the first (period 20212022) and the second Stage (period 2023-2024) of Horizon Europe and key performance indicators Section 8 Roadmap A recap of the main points and conclusions Section 9 Conclusions Summaries of related documents that, together with this SRIA, define the structure, aims and work of EOSC Appendix A Table 1.1: How to read this document - 18 - A note on Terminology Readers are reminded that throughout the document, ‘data’ is used as an encompassing term referring to all digital research artefacts, including datasets, metadata, publications, intermediate results, workflows, notebooks and software code. Similarly, ‘science’ refers to all branches of knowledge and areas of study and research, including arts and humanities subjects rather than in contradistinction to them, while ‘scientist’ refers to all researchers, academics and practitioners in all domains. - 19 - 1 New ways of science The current, digital age is the most recent stage in an evolving continuum of ways in which technology has supported and enhanced science. This section outlines the history of the digitisation of research, establishing the technological context from which the European Open Science Cloud (EOSC) has evolved. It includes lessons to be learned, and developments from which EOSC will benefit and to which it will contribute. 1.1. The opportunity for change 1.1.1. Research in the digital age In the digital age, the world has become instrumented, interconnected and intelligent. Instrumented refers to the fact that digital information is now collected everywhere on the planet using small devices as well as large equipment. Interconnected refers to the fact that digital information produced anywhere on the planet can be made available anywhere else. Intelligent refers to the fact that people and machines can then process this information for the benefit of society at large. In a world that is instrumented, interconnected and intelligent, human activities can be improved by discovering, retrieving, analysing, assembling and computing information in order to extract the knowledge necessary to address challenges at all scales. Among all human activities, research plays an enabling role by producing scientific results that can be exploited by society to address global as well as local problems. Scientific results include publications, data, software and any research artefacts or intermediary results produced during the research lifecycle. The digital age allows the ways research is conducted to change in multiple directions, resulting in better science, increased trust in science, and the ability to meet global challenges. ● Scientists will be able to do better research by getting early (sometimes real-time) access to scientific results, optimising their own work. Disciplines organised around large shared equipment already provide examples of the benefits of sharing information across the globe. ● In a world that is becoming more and more complex, the availability of multiple information sources will allow trust in scientific results to be strengthened by facilitating reproduction of scientific experiments and comparison of outcomes. Trust in science has to become the foundation of the new societal paradigm if Europe wants to maintain and develop its way of life. ● Scientists will be able to engage in multi-disciplinary initiatives to address the key global challenges of the twenty-first century such as climate change, health, food and biodiversity or building energy-efficient vehicles and smarter cities. More generally, all efforts dedicated to achieving the 17 Sustainable Development Goals of the United Nations [UN_SDG] would benefit from access to a wide set of information coming from very different origins. However, while an instrumented, interconnected and intelligent world has unprecedented potential to solve the key challenges of the time, this potential will only be realised if research infrastructures are evolving to allow scientists to make the best use of the available information. - 20 - The European Open Science Cloud (EOSC) will deliver Europe’s contribution to enabling scientists to realise their potential in the digital age. 1.1.2. European leadership When the Horizon Europe programme began in 2021 [Horizon_Europe], Europe has been well placed to lead the world in exploiting digital capabilities at the service of science. After three years of preparation, the initial phase of EOSC was launched in November 2018. More than 35 research and innovation projects have developed foundational technologies and initial services on which Europe can now build. These efforts have also allowed the establishment of a Europe-wide community that is ready to engage further. While other regions in the world have launched their own efforts, none of them have done it at the scale on which Europe has invested. Pursuing the effort to get EOSC operational as part of the Horizon Europe programme will enhance Europe’s leadership position. Through coordination and concentration European research and innovation investments will be more efficient, will be able to address key global challenges and will strengthen the trust in science that society needs to build a common future. 1.2. The request for change 1.2.1. From Gutenberg to Berners-Lee The current way of sharing research was built upon the emergence of the printing process. During the seventeenth century, the first research journals were conceived by academies of sciences. The Philosophical Transactions of the Royal Society was launched in 1665 and received, over the years, articles from scientists such as Newton, Faraday and Darwin. Since then, the publications process has developed in volume but the principles of their use have remained largely identical. Articles are published in journals. Journals are acquired by libraries. Scientists visit their libraries to access the knowledge delivered by their predecessors and colleagues. The digital age has the potential to revolutionise communication between scientists. While peer-reviewed publications remain the ‘official’ way to deliver conclusions (potentially using the internet for faster dissemination and transitioning to an open access business model), many other types of information can be made available, increasing the bandwidth of knowledge sharing. Data, software, intermediate results, workflows and notebooks are often stored in digital form. It is up to the scientists and/or the organisations they work for and/or the organisations that fund their research to decide whether this information should be shared, and how widely. Early and open accessibility of such digital assets form a large part of the transition towards what is now called Open Science. Many researchers will recognise that Open Science is improving science as a whole. However, to date only a subset is convinced that the opportunities it affords to them individually are greater than the drawbacks, while many leaders within the research community argue that data, software and other research artefacts should be kept closed as they are assets that research teams ought to keep for themselves if they want to stay competitive. To change this, establishing a new paradigm for rewards and recognition is essential: it can no longer be based on publications alone. - 21 - 1.2.2. Lindau Declaration Once every year, around 30 to 40 Nobel laureates convene in Lindau, Germany, to meet the next generation of leading scientists: 600 undergraduates, PhD students, and post-doc researchers from all over the world. The Lindau Nobel Laureate Meetings foster exchanges among scientists from different generations, cultures and disciplines. Elizabeth Blackburn is a 2009 Nobel Laureate in Physiology or Medicine for her work in molecular biology. During the 68th Lindau Meeting in 2018, she introduced ten goals for science which subsequently became the core of a 2020 Lindau Declaration [Lindau_Dec]: ● Adopt an ethical code; ● Cooperate globally on global problems; ● Share knowledge; ● Publish results Open Access; ● Publish data in repositories; ● Work transparently and truthfully; ● Change reward system; ● Support talent worldwide; ● Communicate to society; ● Engage in education. Since its original proposal, the Declaration has been open for debate, changes and amendments. The appeal aims to get widespread support for a new approach to global, sustainable, cooperative Open Science. “Around the world, the Open Science movement continues to gain support from a growing number of researchers and members of the public – a scientific meta topic also represented in the Lindau Meeting programme in 2021.” 3 This exemplary initiative illustrates the current status of Open Science. Thought leaders have understood the potential of the digital age, the impacts on the ways to do research and the benefits for society at large. The request for change now comes from the pioneering research community at its most talented level. The fact that developing such a Declaration is needed also shows that strong initiatives have to be taken in order to fulfil the potential and overcome the caution or conservatism of other members of the research communities. 1.3. Open Science Assembling different contributions, Wikipedia defines Open Science as ‘the movement to make scientific research (including publications, data, physical samples, and software) and its dissemination accessible to all levels of an inquiring society, amateur or professional’. It continues: ‘Open science is transparent and accessible knowledge that is shared and developed through collaborative networks. It encompasses practices such as publishing open research, campaigning for open access, encouraging scientists to practice open-notebook science, and generally making it easier to publish and communicate scientific knowledge.’ [Wikipedia_OS] The resulting research activity flows are shown in Figure 1.1. Another definition of Open Science is provided by the FOSTER portal: ‘Open Science is about extending the principles of openness to the whole research cycle […], fostering sharing and 3 Quote from the Annual Report 2021 of the Lindau Nobel Laureate Meetings, available at: https://www.lindau-nobel.org/wp-content/uploads/2021/11/2021_Annual-Report_Web.pdf - 22 - collaboration as early as possible thus entailing a systemic change to the way science and research is done.’ [FOSTER_OS] Figure 1.1: Research activity flows 1.3.1. Brief historical context Until the emergence of academies and journals, science was mostly an individual endeavour supported by patrons. Results were kept secret as much as possible in order for the patrons to be able to benefit from the research results. In the seventeenth century, both the creation of academies where scientists could cooperate and exchange knowledge, and the deployment of printing capabilities, which produced academic journals, allowed a move towards a more open way of science. Nowadays, academic journals have taken a key role in the research lifecycle by allowing the transfer of knowledge, but also as a basis for research assessment through citation mechanisms. When World War II ended, the global scientific community had the opportunity to look at the future with new eyes and with new goals in sight. In the United States, Vannevar Bush delivered the report ‘Science the Endless Frontier’, at the request of President Roosevelt. This report led to the creation of the National Science Foundation (NSF). Public investment in research was recognised as a priority. Since then, public-funded research has developed around the whole world. In the same period, Europe organised cooperation by establishing research organisations such as CERN, for example, which was created in 1954. In the 1980s, with the final objective of defining and implementing an overall development, research and demonstration strategy at Community level, the European Commission established the First Framework Programme covering three years from 1984 to 1987. The total budget dedicated to the programme was €3.75 billion. The programme focused on specific scientific and technical objectives, such as ‘improvement of the management of - 23 - energy resources’; ‘promotion of industrial competitiveness’; ‘improvement of living and working conditions’; ‘promotion of the agricultural competitiveness’; ‘improvement of raw materials management’; ‘stepping up development aid’ and ‘improving the effectiveness of the Community’s scientific and technical potential.’ [EC_FP1] Successive framework programmes came with increasing budgets. With increased funding, over time, countries became engaged in policy decisions regarding the use and impact of research activities. The way knowledge, specifically that created with the support of public funding, would be shared became a key societal and political topic. The debate was fuelled by prior research. For example, the Mertonian paradigm, introduced by Robert Merton in his book The Sociology of Science in 1942, was based upon four ‘norms’: ● Communism. All scientists should have common ownership of scientific goods (intellectual property), to promote collective collaboration; secrecy is the opposite of this norm. ● Universalism: Scientific validity is independent of the socio-political status/personal attributes of its participants. ● Disinterestedness. Scientific institutions act for the benefit of a common scientific enterprise, rather than for the personal gain of individuals within them. ● Organised scepticism. Scientific claims should be exposed to critical scrutiny before being accepted, both in methodology and institutional codes of conduct. At the turn of the twenty-first century, the digital age created new avenues for knowledge sharing. These new opportunities have been recognised by research communities across the world. More international collaborations were launched, leveraging the interconnections made possible by the internet. Open Science emerged from the meeting of the needs (sharing knowledge) with the means (digital technologies). Figure 1.2: Open Science taxonomy (from the FOSTER project) - 24 - The European Commission identified early the potential of digital technologies in changing the way research is conducted. In 2017, the FOSTER project was funded to study the practical implementation of Open Science in Horizon 2020 and beyond [FOSTER]. The project developed the FOSTER portal as a platform that brings together the best resources addressed to those who need to know more about Open Science, or need to develop strategies and skills for implementing Open Science practices in their daily work. The resources are structured around the Open Science taxonomy shown in Figure 1.2 above; the role of Open Science in the research lifecycle is summarised in Figure 1.3 below. Figure 1.3: Research lifecycle and Open Science (from the FOSTER project) 1.3.2. Open Science facets: documents, data and software The FOSTER project also identified the Open Science ‘facets’ that could be shared by scientists within and between research communities (Figure 1.4). - 25 - Figure 1.4: Open Science facets (from the FOSTER project) Those ‘facets’ are different in nature and therefore sharing them requires specific approaches. The complementarity and differences between documents, data and software are well known by the computer science community. The sharing processes for each of these are explored in the following sections. 1.3.2.1. Documents Publications, notebooks and educational materials are documents written in natural languages. They are designed to be read by people, while machines may leverage their content through document processing. Publications were (and still are) the basis for information exchange between scientists. The first instance of the World Wide Web to be deployed, in the early 90s, was a Web of documents. Therefore, technology met user needs and digital publications became the norm. This soon created friction with regard to the intellectual property rights, which up to this point were mostly owned by publishing corporations. On 14 February, 2002, the Budapest Open Access Initiative produced its original declaration which started as follows: ‘An old tradition and a new technology have converged to make possible an unprecedented public good. The old tradition is the willingness of scientists and scholars to publish the fruits of their research in scholarly journals without payment, for the sake of inquiry and knowledge. The new technology is the internet. The public good they make possible is the world-wide electronic distribution of the peer-reviewed journal literature and completely free and unrestricted access to it by all scientists, scholars, teachers, students, and other curious minds.’ [Budapest_OAI] The Open Access movement launched the debate with publishers which is now focused on legal or contractual issues about ‘ownership’ of the content. Since then, multiple examples of open access initiatives have flourished around the world. Launched in the United States in 1991 and currently managed by the University of Cornell, arXiv is an open access repository of electronic preprints (known as e-prints) approved for posting after moderation, but not full peer review. It consists of scientific papers in the fields of mathematics, physics, astronomy, electrical engineering, computer science, quantitative biology, statistics, mathematical finance and economics, which can be accessed online. In many fields of mathematics and physics, almost all scientific papers are self-archived on the arXiv repository before publication in a peer-reviewed journal. Following the arXiv model, similar archives have been established in many different disciplines. Recently, preprints have become popular in life sciences and have turned out to be essential in the scientific communication related to COVID-19. - 32 - Federating security policies implemented by the research infrastructures may, in specific cases, limit openness. EOSC will have to strike the right balance between trustworthy security policies and open access to research artefacts. There are other cases where security will have to be taken into account. These relate to the content of the information itself, which may require special attention. It will be the role of the infrastructure governance to decide whether or not information may be made available openly and to whom. 1.3.4.3. Property Data acquisition can be a costly process. The entity performing that process has rights and responsibilities with regard to defining the use of the data. Depending upon the funding mechanism, the ‘owner’ may decide to limit access to the data. Open Science infrastructures will have to provide a way for stakeholders to exercise their rights, possibly limiting open access to the data. In any case, the stakeholder policy will have to be shared openly. 1.3.4.4. Sovereignty Member States may decide that data management has to follow specific rules. Member States should be able to exercise their full right and power over documents, data and software, limiting full openness. Striking the right balance between ensuring sovereignty and ease of use will require special attention. As a consequence of those limits, an infrastructure for Open Science has to offer capabilities for identity and rights management. Being a federation of research infrastructures, those capabilities will have to be powerful enough to offer individuals, organisations or governments a way of exercising the required control while keeping knowledge ‘as open as possible, as closed as necessary’ 5 . 1.4. Next-generation infrastructure 1.4.1. Learning lessons from the recent past In order to position EOSC in its context, it is important to briefly review the evolution of digital services made available to scientists over the years, together with the key technologies (networking, hardware and software) that were developed and deployed to allow those services to flourish, as well as the funding models and the policy decisions that exploited these new capabilities. This review is summarised in Table 1.1 and allows EOSC to be positioned in time and technological space, leading to the definition of the ‘raison d’être’ of the initiative. 1.4.1.1. 1970 The internet, mainframes and leased lines The world recently celebrated 50 years of the internet. In September 1969, a few characters were exchanged between four mainframe computers installed in different locations in the west of the United States. The first message was intended to send the word login from one computer to another (and it failed). It was the first step towards delivering the remote login service: allowing an end user to use remote computers such as the mainframes or supercomputers of the day. 5 For research artefacts, the interpretation of “as closed as necessary” is “as restricted as necessary”. - 33 - In the years following this historical moment, the internet allowed the development of many other services, predecessors to the ones routinely used today: email, file transfer, chat. For a long while those services were only deployed within research communities. It is important to note that this deployment was rapidly global. Connections, gateways and routers were assembled to build the first generation of the global internet infrastructure and to allow scientists to improve the way they collaborated. Funding mostly came from research-funding - 34 - organisations. The development of internet Standards grew as a bottom-up effort, driven by the Internet Engineering Task Force (IETF), exploiting the communication capabilities provided by the internet (e.g. mailing lists, file transfers, news) to assemble hundreds of computing and telecommunications scientists and design the internet architecture. Table 1.1: EOSC in its technological context - 35 - 1.4.1.2. 1980 Unix, the personal computer, ethernet A few years later, progress in microprocessor technologies allowed the design of personal computers, which rapidly became the main tool for scientists to do their research. Scientific workstations were born, providing researchers with the best technologies of the day in terms of computing, graphics and networking. Equipped with office automation and computer-aided design software, these devices were connected to local area networks, changing the way scientists would collaborate within a team or a laboratory. Unix, C, C++, TCP-IP and X Window System were the software standards that allowed interoperability between these devices. Scientists were able to share their results within their teams and/or their laboratories by sharing databases, for example. Distributed file systems allowed the development of new ways of collaboration. It is important to note that many of those software standards were developed within computer science laboratories both private and public (Unix, C and C++ at Bell Labs, TCP-IP at UC Berkeley, X Window System at MIT). 1.4.1.3. 1990 The World Wide Web: the internet becomes pervasive Twenty years after the birth of the internet, the Web was invented at CERN, the European Organisation for Nuclear Research. In March 1989, Tim Berners-Lee wrote a memo entitled ‘Information Management: A proposal’. The project was approved and Berners-Lee developed the World Wide Web using a NeXT machine, the most advanced workstation of the moment. It is interesting to note that user research organisations were also at the origin of the deployment of the Web. For example, in the United States, the first Web server was installed at Stanford Linear Accelerator Center (SLAC). Also, while many Web browsers blossomed around the world, Mosaic from the National Center for Supercomputing Applications was made available on PC, Mac and Unix machines and became a huge success. Scientists were able to share documents, graphics and images thanks to the worldwide deployment of the Web. Essential generic services such as discovery or service catalogues were developed within computer science departments (Stanford University). Open source efforts delivered key software components of the infrastructure such as Apache, the Web server, wiki, the collaborative tool that was to be used widely by research communities. Services that were not planned in the original design of the internet (e.g. real-time signals) and that used to require specific networks (telephone, television) moved to the internet and offered new opportunities for innovation. As a whole, the success of the Web fuelled the massive deployment of the internet infrastructure with private and public funding. In order to ‘lead the Web to its full potential’, Tim Berners-Lee moved from CERN to MIT and launched the World Wide Web Consortium (W3C), following the lessons learned from the X Window Consortium. By design W3C would have multiple hosts and Inria, the French National Institute for Research in Digital Science and Technology, in Europe and Keio University in Japan became the European and Asian hosts of W3C. 1.4.1.4. 2000 Documents and data While the first version of the World Wide Web allowed the creation of a Web of documents, using its original components URL, HTTP and HTML, very soon the request for a Web of data led W3C to develop XML and a family of related standards. This effort brought together different communities with participants coming from academia, public and private organisations. It also became clear that beyond the description of syntaxes, in order to make full use of data, semantics needed to be formalised. Tim Berners-Lee launched the development of the Semantic Web within W3C. The growing impact of the Web on society was recognised by W3C. The Web Accessibility Initiative (WAI) became part of the strategic - 36 - priorities of the consortium in order to design a Web that could be used by people with disabilities. At the same time, in order to strike the right balance between public and private investments, W3C developed a royalty-free policy for using W3C standards. During the same period, the deployment of the internet was able to benefit from a wide range of new networking technologies, from fibre optics (within the core of the network) to Wi-Fi (at the edge). The last-mile challenge addressed by ADSL technology was also about to be covered, by the deployment of mobile infrastructures. The design of the internet allowed the use of all these technologies in order to build the resilient infrastructure it is today. The vision of Tim Berners-Lee became true when ‘thousands of flowers bloomed’ on the Web [Forbes_TB-L], ranging from an open encyclopaedia to the emergence of social networks. The internet infrastructure was able to carry telephone and television signals. Digital photography was about to become widespread. 1.4.1.5. 2010 Smartphones, cloud computing, linked data platform The momentum of the internet only accelerated further when microelectronics technologies allowed the functions of a telephone, a computer and, soon enough, a television to be embedded in a handheld device. The smartphone was born, filling the pockets of millions of people around the globe, including researchers. At the same time, the decreasing cost of computing and storage resources and the improvements in bandwidth of the internet allowed the launch of the cloud computing paradigm. Progress in computing architecture during the early 2000s allowed the Grid Computing route to be added to the general evolution of supercomputing. Scientific problems could be addressed with a wide range of architecture possibilities. Building upon the outcome of the Semantic Web efforts, W3C launched the Linked Data Platform initiative with the goal of creating the architecture components that will allow data to be ‘linked’ and lay the ground for the internet of FAIR (findable, accessible, interoperable and reusable) data. Many people in the research community recognised that the time had come to leverage the progress of the internet infrastructure, leading to the launch of the Research Data Alliance (RDA) in 2013 to ‘build the social and technical bridges to enable the open sharing and reuse of research data’. 1.4.1.6. 2020 Lessons learned The emergence, over the years, of so many new digital products and services followed similar paths: ● New user needs served by breakthrough technologies; ● Next-generation services deployed on existing infrastructures; ● Use of novel services pioneered by research communities and then deployed for the general public; ● Initial efforts supported by public funds and then embraced and further developed by industry; ● Innovation fuelled by private funds; ● Pervasive deployment delivered by open and proprietary offerings. - 37 - Over the last 50 years, exceptional developments have allowed scientists to use machines that improve the exchange of documents, data, software and related information between people. Looking into the future, further improvements in digital technologies will create new opportunities. Machines will be assembled into complex systems and put at the service of research teams composed of experts from any discipline working from anywhere in the world. While the potential offered by current (and future) devices is unique in human history, the current limits come from the programmability of those complex systems to develop friendly user-oriented services and the capacity to find, access and reuse data in an interoperable framework. The Horizon Europe programme will address many of these challenges through various partnerships and the EOSC Partnership will form links with those inside Europe. 1.4.2. Networking: the next-generation internet (NGI) The NGI is an ambitious research and innovation programme with an EC investment of more than €250m for the initial phase between 2018 and 2020, and is an important part of the upcoming Horizon Europe programme (2021–2027). Focus has been on advanced technology applied to evolve the internet into an ‘Internet for Humans’. The initiative addresses the challenges of privacy and trust, search and discovery, by promoting decentralised architectures, blockchain, the Internet of Things (IoT), social media and interactive technologies, as well as technologies supporting multilingualism and accessibility. Also, the whole new area of next-generation Internet of Things research will be covered under the NGI programme. EOSC will benefit from the evolution of the internet towards an ‘Internet for Humans’. EOSC will be able to face the challenges of privacy, security, property and sovereignty by leveraging the results of the NGI initiative. EOSC will also make use directly of the new IoT technologies and infrastructures, as this is one of the sources of the large amount of data that can be used for research inside the EOSC ecosystem. 1.4.3. Hardware: the computing continuum In their paper ‘Harnessing the Computing Continuum for Programming Our World’, Beckman, Beck, Dongarra et al. describe the challenges facing scientists in mastering systems composed of elements as different as smart sensors at the one end and supercomputers at the other [Beckman_2019]. The Computing Continuum is described in Figure 1.6. Figure 1.6: The Computing Continuum (from the ETP4HPC SRA) - 38 - In its Strategic Research Agenda (SRA) published in March 2020, the Institutionalised Partnership EuroHPC extends the concept and introduces a new paradigm called the ‘Digital Continuum’: ‘The rapid proliferation of digital data generators, the unprecedented growth in the volume and diversity of the data they generate, and the intense evolution of the methods for analysing and using that data are radically reshaping the landscape of scientific computing. The most critical problems involve logistics of wide-area, multistage workflows that move back and forth across the computing continuum, between the multitude of distributed sensors, instruments and other devices at the network’s edge and the centralised resources of commercial clouds and HPC centres.’ [ETP4HPC_SRA] The ETP4HPC SRA has been designed to strengthen and develop further the European position with respect to the ‘Digital Continuum’ during the Horizon Europe programme. EOSC will ensure close collaboration in order to contribute to the Digital Continuum. 1.4.4. Software: visualise, analyse, predict Key elements of the research lifecycle involve observation, explanation and prediction. If and when large datasets are available, scientists need to use machines to support their work. Observation requires machines to help in visualisation; explanation requires machines to analyse data and derive models; prediction requires machines to check hypotheses. The larger the datasets, the harder becomes visualisation, analysis and prediction. Scientists need to use advanced software in order to improve their insights. Scientists also use machines and software to check hypotheses, simulate phenomena and strengthen their ideas and models. During the Horizon Europe programme, the AI (Artificial Intelligence), Data & Robotics partnership will help position Europe in the global development of AI technologies. The summary of the partnership proposal states: ‘Access to relevant and high-quality data is widely recognised to be one of the crucial elements in building an AI economy in Europe. Building on the great efforts to make industrial and public sector data more accessible during Horizon 2020, the access to data will have to scale up in Horizon Europe, address a broader set of sectors and drastically increase the quantity of high-quality datasets available.’ [EP_AID&R] EOSC will make available the high-quality scientific datasets to be consumed by machinedriven AI applications at the service of science. 1.4.5. Data: findable, accessible, interoperable and reusable The first Web client, developed by Tim Berners-Lee, was both a browser and an editor. A user could therefore not only read but could also create content. The Web was conceived originally as a collaborative space. However, when the first popular browser, Mosaic, came along, in 1993, it included images but the editing capability was taken out. It was considered too difficult a problem. The Web was also originally designed to be a space for data as well as documents. The Linked Data Platform, recently developed, is an important step towards giving data first-class citizen status on the Web. These two principles, of read-write capability and managed data accessibility, were part of the original vision for the Web. They are still not available as Web features. They may be - 39 - present at the application level. For example, wikis or social networks offer the write capability within their own environment. Since 2015, Tim Berners-Lee has been working on SOcial LInked Data (SOLID) in order to offer those two capabilities for the whole Web. Those capabilities are essential in order for EOSC to achieve its full potential. Referring to Neil Armstrong’s famous sentence when landing on the moon, Berners-Lee defines what he thinks is ‘A small step for the Web’: ‘I have always believed the Web is for everyone. […] This is why I have, over recent years, been working with a few people at MIT and elsewhere to develop SOLID, an open source project to restore the power and agency of individuals on the web. […] SOLID is a platform built using the existing web. It gives every user a choice about where data is stored, which specific people and groups can access select elements, and which applications you use […] SOLID is guided by the principle of “personal empowerment through data” […] I’m incredibly optimistic for this next era of the web […] The future is still much bigger than the past.’ [TB-L_Step] ‘The issue with writing data, as Wikipedia and others have learned, is that you need a degree of control over who can write what. The writer needs to have permissions describing what individuals can do to the data. And to have permissions you need to have a system for identity – a way of uniquely confirming that an individual is who they purport to be. Hence, based on existing Web standards and the result of decades of work, SOLID has read-write functionality, incorporating permissions and identity, along with data manageability and real-time updates. It realises the Web as originally envisioned and provides a platform for the next generation of truly empowering and innovative applications.’ While the success and the deployment of SOLID is yet to be proven, the issues that SOLID addresses are at the core of what EOSC needs in order for scientists to find, access and reuse interoperable research results. 1.4.6. Machines for scientists: EOSC foundations ‘Machines need direction from human minds, and human minds need inspiration from human leaders’. Arno Penzias, Nobel Prize-winning physicist, reminds us that it is up to us to build the environment and the infrastructures that will facilitate the exchange and composition of ideas, allowing scientists to cooperate globally and help solve the scientific and societal challenges of our time. In order for scientists to share the universe of scientific networked-accessible information, the essential foundations are: ● Persistent identifiers: a mechanism for naming and locating documents, data and software in a persistent manner; ● Metadata and ontologies: a mechanism for discovery of and access to documents, data and software in a structured manner; ● Internet identity: an authentication and authorisation infrastructure (AAI). The first mission of EOSC will be to provide those mechanisms and that infrastructure to enable machines to get direction from human minds for the benefit of all. EOSC will allow identified scientists to store, share, discover and access identifiable documents, data and software. - 40 - EOSC will allow identified scientists to (re)use identifiable documents, data and software, exploit identified services, reproduce experiments and address the problems of our time. Section 1 of this SRIA has presented an overview of the development of scientific practice, as influenced by changing ideas and evolving technologies. Section 2 provides further context, focusing on the recent, current and future status of and strategy towards science and data in Europe, and the evolution of EOSC. - 41 - 2 EOSC in the making The European Open Science Cloud (EOSC) initiative is the tangible outcome of a number of key European and global policy milestones and position statements regarding Open Science. EOSC is an integral part of, and supports, the European Commission’s strategy for realising the European Research Area (ERA), in particular the policy priorities of Open Innovation, Open Science and Open to the World and the goal of findable, accessible, interoperable and reusable (FAIR) data. This section outlines the EC policy context for and stages of EOSC’s development, including its governance structure and activities during the transition period 2019–2020, and the landscape of national infrastructures and international initiatives, as well as the role of the governance of EOSC in the form of the EOSC Association. 2.1. European Research Area The European Research Area (ERA) was launched by the European Commission in 2000 with the aim of better organising and integrating Europe’s research and innovation systems and enhancing cooperation between the EU, the Member States, their regions and their stakeholders. It also aimed for the free circulation of researchers, scientific knowledge and technology throughout the EU and focused on stimulating cross-border cooperation and on improving and coordinating the research and innovation policies and programmes of the Member States. In May 2016, the Commission published ‘Open Innovation, Open Science and Open to the World – a vision for Europe’ [EC_Open_Vision] as a key policy priority for realising the ERA, with the following goals in mind: ● Open Innovation will help Europe capitalise socially and economically on research and innovation results by bringing more actors and investments into the research and innovation process; ● Open Science will help Europe benefit from digitisation and support new ways of doing research and innovation, as well as opening up access to research data and results via digital technologies and collaborative tools; ● Open to the World will make Europe a leading voice in global debates and tackle societal challenges by engaging more in science diplomacy and global scientific collaboration. It is as part of this strategy for Open Science that the European Commission adopted the European Cloud Initiative – Building a competitive data and knowledge economy in Europe [EC_Cloud] and launched the initiative of creating the European Open Science Cloud (EOSC). Both initiatives were designed to give a strong push in Europe towards Open Science and findable, accessible, interoperable and reusable (FAIR) research data management and to ensure that European researchers and professionals reap the full benefits of data-driven science. Building EOSC basically equates to designing a virtual commons where science producers and science consumers come together for more insights, new ideas and more innovation. In April 2020, the European Commission introduced the ERAvsCorona Action Plan as part of the EU response to the coronavirus pandemic [EC_ERAvsCoronaAP]. Building on the overall objectives and the tools of the European Research Area, the action plan is a working document developed jointly by the Commission and national governments. It covers short-term actions based on close coordination, cooperation, data sharing and shared funding efforts. - 48 - Use Case: COVID-19 – understanding the SARS-CoV-2 virus, its structure and societal impact The ongoing COVID-19 pandemic has a profound impact on Europe’s citizens. Faced with an unknown, dangerous and rapidly spreading pathogen, a massive research response was mounted to understand the novel virus and to develop clinical diagnoses and prognoses, treatments, preventive measures and vaccines. The impact of the pandemic has been felt in all aspects of society, and understanding these effects is of major importance for future pandemic preparedness. Whereas the focus has been on biological, clinical and epidemiological data, it is evident that research will enter a second phase with emphasis on social behaviour, economic measures and the effectiveness of vaccination programmes. Another aspect is how to improve information sharing and filter out fake news. For example, the effects of social media on information access and society response remain poorly understood, but disinformation – targeted and incidental – has been rife throughout the pandemic. There is a need to combine and connect biological, genetic, epidemiological, health, social, political and economic data to monitor COVID-19. European researchers contributing to the response to the COVID-19 pandemic thus need to be able to store, share, access, analyse and process research data and other research digital objects across disciplines and national borders and to collaborate with global partners. Therefore, instead of setting up new data silos, the European COVID-19 Data Portal [C19_Portal] should be complemented with social and life sciences data to promote an integrated understanding of the European outbreak response and future preparedness, as well as to demonstrate the value of FAIR data to society and public engagement. The integration and management of health data is an enormous challenge that will have an unprecedented impact on personalised health. Viral and human infectious disease data on SARS-CoV-2 and COVID-19 from national centres as well as clinical data should be linked across research fields, e.g. with contextual economic, social, cultural and migration data from the social sciences and humanities and in compliance with the security and protection of sensitive data collection and analysis. To bring together these data, the question of how to make the metadata of data objects interoperable needs to be addressed. Within the science cluster projects (ENVRIFAIR, EOSC-Life, ESCAPE, PaNOSC, SSHOC), there are different metadata schemas in use based on strong and well-established domain standards (e.g. ECRIN metadata schema for clinical research, MIABIS for biobanks, DICOM for images, DDI for social sciences). The goal is to assess the metadata schemas and domain catalogues and to develop a strategy for how to map and relate the different schemas to support the search for data objects related to COVID-19. Within EOSC, the science clusters will review models for automatic raw and intermediate data preservation and sharing, including software and analysis workflows; identify and share best practice; and link the catalogues with the emerging EOSC services and the EOSC Interoperability Framework. This work will align with, and enrich, the data currently being exposed within the European COVID-19 Data Portal. All science clusters will bring their experience and expertise to organise data from many different sources (including addressing multilingualism issues) and different types and modalities, so that they can be easily gathered, analysed and modelled holistically, and not remain fragmented as at present. These combined analyses are still rare, due to lack of appropriate platforms, infrastructures and FAIR data. On one side, it is important to ensure the fulfilment of the ethical and legal boundaries (‘as open as - 49 - possible, as closed as necessary’9), although mechanisms for pseudonymisation and traceability do exist, enabling data sharing. On the other side, streaming and imaging data will require vast amounts of storage, network and computing capacity, and the same goes for multilevel social network analyses. The growing need for data storage and the large requirement of network bandwidth for sharing and analysis is a fundamental issue in advanced research across all domains which EOSC aims to address. 2.5. A history of EOSC Open Science has been a policy priority of the European Commission since 2016 [EC_Open_Vision]. Together with Open Innovation, which will involve public and private sector actors in research to create new tools and services, and Open to the World, which will ensure involvement and open collaboration with non-European stakeholders, Open Science will open up the whole research process through digital technology. Open Science is a transformative driver that will shape the research and innovation policies for a renewed European Research Area. To further develop and implement the policies for Open Science, the European Open Science Policy Platform (OSPP) [EC_OSPP] was established as an advisory group consisting of stakeholders from the research community. The OSPP issued its final recommendations in 2020. To enable the development and uptake of Open Science in Europe, the EC has proposed the creation of a European Open Science Cloud (EOSC). EOSC will essentially involve the federation of existing research data infrastructures and the realisation of a Web of FAIR Data and Related Services for Science, making research data interoperable and machine-actionable following the FAIR guiding principles [FAIR_Principles]. This web of data will allow researchers to find, exploit and combine linked datasets, providing a basis for artificial intelligence (AI) tools, leading to new discoveries and research paradigms. EOSC will initially focus on traditional research data but will also include research publications and research code. EOSC will encourage FAIR datasets to be made fully open, and will follow the principle of ‘as open as possible, as closed as necessary’ 10 . This is typically important for biomedical, military, sensitive, private and commercial datasets which may not be opened immediately or fully or indeed ever released. In an initial phase of development from 2017 to 2020, the EC made a financial investment of approximately €320 million to begin building the foundations of EOSC through project calls in Work Programmes in Horizon 2020. This investment was targeted to develop a new panEuropean access mechanism to public e-infrastructures, to coordinate related national activities, to connect European research infrastructures (RIs) to EOSC, to set up and begin the implementation of the FAIR guiding principles, and to start a FAIR-compliant certification scheme for research data infrastructures. These projects have involved the community of stakeholders of EOSC and have been steadily developing the broader EOSC ecosystem. 11 To help steer the initial development of EOSC, the EC appointed two high-level Expert Groups, which delivered recommendations on a vision for EOSC in 2016 [EC_EG1_EOSC] and on how to practically implement EOSC in 2018 [EC_EG2_EOSC], and an Expert Group on FAIR data, which offered recommendations on how to make FAIR a reality in 2018 [EC_EG_FAIR]. 9 For research artefacts, the interpretation of “as closed as necessary” is “as restricted as necessary”. 10 For research artefacts, the interpretation of “as closed as necessary” is “as restricted as necessary”. 11 See Annex 1 of the landscape report for a list of EOSC projects funded under Horizon 2020 [EOSC_Landscape]. - 50 - The initial development phase supported more than 35 projects, laying the foundations of EOSC and showcasing its diversity and complexity. The EOSCpilot project engaged extensively with stakeholders and proposed a governance framework and policies, as well as developing interoperability pilots across scientific domains [EOSCpilot]. EOSC-hub brought together service providers to create a single contact point to discover, access and use a wide range of resources for data-driven research [EOSC-hub]. The five ongoing cluster projects will connect the European Strategy Forum on Research Infrastructures (ESFRI) projects and landmarks to EOSC in the domains of environmental sciences via ENVRI-FAIR [ENVRI-FAIR], life sciences via EOSC-Life [EOSC-Life], astronomy and particle physics via ESCAPE [ESCAPE], photon and neutron sciences via PaNOSC [PaNOSC], and social sciences and humanities via SSHOC [SSHOC]. The five regional projects aim to coordinate the efforts of national and thematic initiatives in contributing to EOSC through groupings of European countries via EOSC-Nordic [EOSC-Nordic], EOSC-Pillar [EOSC-Pillar], EOSC-Synergy [EOSC-Synergy], ExPaNDS [ExPaNDS] and NI4OS-Europe [NI4OS-Europe]. HNSciCloud established a hybrid cloud platform to support high-performance and big-data computing through commercial procurement [HNSciCloud], work that is continuing through ARCHIVER [ARCHIVER] and OCRE [OCRE]. Finally, towards the end of Horizon 2020, a number of projects were awarded, i.e. EOSC Future [EOSC_Future], EGI-ACE [EGI_ACE], DICE [DICE], RELIANCE [RELIANCE], OpenAIRE-Nexus [OpenAIRE_Nexus], and C-SCALE [C_SCALE]. In 2022, nine projects were awarded under the first Horizon Europe INFRAEOSC call. Some started their work as of 1 June 2022. As of 1 October 2022, all nine projects, i.e. AI4EOSC [AI4EOSC], EOSC Focus [EOSC_Focus], EOSC4Cancer [EOSC4CANCER], EuroScienceGateway, FAIRCORE4EOSC [FAIRCORE4EOSC], FAIR-EASE, FAIR-IMPACT [FAIR_IMPACT], RAISE, and Skills4EOSC, have started. A first version of the Vademecum 12 was developed detailing the collaboration between the projects and with the EOSC Association, to maximise the impact of the projects for the EOSC ecosystem. From the second Horizon Europe INFRAEOSC call, six projects have been awarded and are in the process of starting, i.e. GraspOS, CRAFT-OA, Blue-Cloud 2026, AquaINFRA, RDA TIGER, and SciLake. 2.5.1. Transition period 2019-2020 The initial phase of development for EOSC is tied to the funding programme of Horizon 2020, which comes to an end in December 2020. To direct the strategic implementation of EOSC, the EC published an implementation roadmap in 2018 detailing six main action lines to realise an architecture, data, services, access and interfaces, rules and governance for EOSC [EOSC_Roadmap]. This roadmap not only served the first implementation phase of EOSC in 2018–2020 under Horizon 2020, but also prepared for the second implementation phase of EOSC under the new funding programme of Horizon Europe for 2021–2027. The roadmap envisioned a pan-European federation of research data infrastructures built around a federating core, providing access to a wide range of publicly funded services supplied at national, regional and institutional levels, and to complementary commercial services. Lessons learned in the first implementation phase have shown that while the project-based 12 The September 2022 version of the Vademecum, presented at the Horizon Europe EOSC-related projects coordination meeting in Brussels, on 30 September 2022, is available at: https://bit.ly/EOSCvademecum22 - 51 - approach is very successful in involving the many stakeholders and communities in developing the EOSC ecosystem, the individuality and freedom of projects has led to a fragmented landscape of systems and stakeholders. With the aim of bringing the community together and ensuring a smooth transition from the first to the second implementation phase of EOSC, a three-tiered transition governance structure was established to run from 2019–2020 [EOSC_Gov]. The EOSC Executive Board, consisting of eight members representing organisations and three independent experts, advised and supported the strategy, implementation, monitoring and reporting on the implementation progress [EOSC_EB]. The EOSC Governance Board, consisting of representatives of Member States, Associated Countries and the EC, oversaw and supported the activities of the Executive Board and ensured an effective implementation of EOSC [EOSC_GB]. The Stakeholder Forum, consisting of the full EOSC community of organisations, projects and initiatives, allowed the collection of input and provision of feedback on the implementation of EOSC via events, online consultations, and the interactive Liaison Platform [EOSC_SF]. The EOSC governance structure was supported by the EOSCsecretariat.eu project, which not only functioned as the governance secretariat, but also managed a co-creation fund for activities and proposals from the stakeholder community to co-develop and co-implement EOSC [EOSC_Sec]. The Executive Board identified priority areas for EOSC and created six working groups (WGs) consisting of experts from the EOSC projects and stakeholder community [EOSC_WGs]. WG Architecture was defining a technical framework to enable and sustain an evolving EOSC federation of systems, including application programming interfaces (APIs), authentication and authorisation infrastructure (AAI), and persistent identifiers (PIDs) [EOSC_WG_Arch]. WG FAIR was defining requirements for developing, assessing and certifying EOSC services in order to foster cross-disciplinary interoperability through FAIR [EOSC_WG_FAIR]. WG Landscape was mapping the landscape and readiness of existing research infrastructures in Europe that could be connected to EOSC [EOSC_WG_Land]. WG Rules of Participation was designing the rules to define the rights and obligations governing transactions between EOSC users, providers and operators [EOSC_WG_RoP]. WG Skills & Training was providing a framework for a sustainable training infrastructure to support the uptake of EOSC [EOSC_WG_Skills]. Finally, WG Sustainability was providing recommendations on the implementation of a scalable and sustainable EOSC, including business models, integration of national infrastructures, and legal models for EOSC [EOSC_WG_Sustain]. The activities of the Executive Board and WGs were steered by a Strategic Implementation Plan (SIP), which defined the background, vision, priorities and main goals of the Executive Board and WGs for EOSC [EOSC_SIP], and a work plan for 2019–2020, which set out the timeline, methods and delivery of key outputs of the Executive Board and WGs [EOSC_Work_Plan]. The overarching objective of the Executive Board was to provide recommendations on mechanisms and possible forms for governing EOSC in the second phase of implementation in 2021–2027 and to hand over all outputs to the new governance structure. In contrast to the first implementation phase of individual projects independently realising EOSC, the second implementation phase should have consolidated all project outputs and ensured directionality (through a common vision and objectives) and additionality (through complementary commitments and contributions). The transition governance bodies identified a Co-programmed European Partnership as the best instrument to overcome the fragmentation and to provide a framework for collaboration and pooling of resources at European, national, regional, and institutional levels. The motivation by all Member States to - 52 - establish an EOSC Partnership in Horizon Europe was clearly expressed through the EOSC Governance Board since 2019. This reflected a general interest by the Member States to target the whole research ecosystem in Europe and not only the EU tier implemented through calls. Such a partnership would strengthen ownership by the research communities, achieve scale by aggregating demand by researchers and other users, and pool existing capabilities at European, national and regional levels. An EOSC Partnership can be seen as a means to obtain commitments to realise the EOSC-Core and expand it iteratively – possibly with new partners – to realise the wider, trusted and open EOSC distributed environment. The transition governance bodies founded a new legal entity called the EOSC Association which was envisioned to have as members all relevant stakeholders in the EOSC ecosystem and to enter into a contractual arrangement with the European Commission to direct the Partnership under Horizon Europe (further information is provided in Section 2.7). 2.5.2. National infrastructures across Europe The Landscape Working Group established by the EOSC Executive Board set out to survey and document the landscape of infrastructures and initiatives across Europe related to the development of EOSC. The work built on existing surveys and information provided by national authorities, various stakeholder communities and the relevant Horizon 2020 projects in close collaboration with the Member States and Associated Countries. Initial inputs included the recent report of the e-Infrastructure Reflection Group, findings of the EC group of national points of reference, the surveys carried out by the OpenAIRE project, the EOSC-Pillar project, analysis of preliminary mapping of the UK’s research and innovation infrastructure landscape, the experience of the ESFRI workshop on cross-disciplinary collaboration of ESFRI landmarks, other relevant documents identified by the WG members, and outcomes of the survey (country sheets) performed by the Landscape WG itself. The WG had collated inputs from einfrastructures including data and high-performance computing (HPC) facilities, from European and national research networks, from pan-European infrastructures and ESFRI roadmap projects and clusters, and from supporting initiatives such as the Research Data Alliance (RDA). Information has so far been collated on 49 Member States and Associated Countries. The WG had also surveyed the landscape of policy development across Europe regarding Open Science and EOSC. Most countries have research evaluation policies in place, as one would hope. The majority (61%) of the Member States and Associated Countries responding at the time had policies in place regarding open access to scholarly publications, but only 34% had a policy in place regarding FAIR data (though, encouragingly, 44% have one either in planning or under development). Few countries seemed ready to mandate that research data should automatically be made open. Relatively few countries (21%) mentioned EOSC in their policies, but 38% planned to do so in future; only three countries so far (Bulgaria, Denmark, and Romania) included mention of EOSC in their criteria for funding. More than half of the responding countries had nominated contact points for Open Science (53%) and for EOSC (42%). Taken together, sources showed that there was a significant investment across Europe in einfrastructures and data-oriented infrastructures. For EOSC to reach its full potential, these investments needed (and still need) to be either federated as part of EOSC or made accessible to users through EOSC. There has also been a significant, though not yet universal, adoption of policies towards Open Science and FAIR data. Although EOSC has not been very visible up until then, as part of national investment strategies, there was evidence that future policies - 53 - and strategies will increasingly align around the concept. While it was proven hard to obtain definitive and quantitative data on national levels of investment, it was already clear that the bulk of these countries have significant investments in national e-infrastructures of one kind or another that could in principle either be federated as part of EOSC or made accessible to users through EOSC. The same applied to many of the data-intensive ESFRI landmark infrastructures. The scale of such investments over the past decade was already certainly in the billions of euros and hence much larger than the planned central investment in the EOSC core. This showed that EOSC will only reach its true potential through effective federation of national and research infrastructure resources. EOSC as a sustainable collaboration system aims to link research data repositories and interconnect services and infrastructures. Research infrastructures are facilities that provide resources and services for research communities to conduct research and foster innovation. The landscape of European infrastructures was surveyed in this context, mainly focusing on the description of e-infrastructures and research infrastructures across different research areas. The survey of the landscape regarding EOSC-relevant infrastructures covered the following: ● einfrastructures. e-Infrastructures address the needs of European researchers for digital services in terms of networking, computing and data management, and foster the emergence of Open Science [EC_OS] as an essential block of the ERA. ● Networking and other services. Each European country has a National Research and Education Network (NREN), connecting research and higher education institutions with high-performance networks, and offering a range of related services (e.g. eduroam). ● Data infrastructures. Data infrastructures consist of data assets supported by people, processes and technology and include the technical and human infrastructures that support management and sharing of research data. ● Computing infrastructures. The EuroHPC Joint Undertaking [EuroHPC_JU] is a legal and funding entity with the aim of developing a pan-European supercomputing infrastructure and supporting research and innovation activities by developing a European supercomputing ecosystem, stimulating the technology supply industry, and making supercomputing resources in many application areas available to a large number of public and private users. In several countries, EGI and EUDAT coordinate significant high-throughput compute (HTC) and data services at an international level based on a partnership model, while HPC centres generally join the PRACE partnership initiative and participate in EuroHPC. EOSC will bridge this separation and help address the question of the relation between centralised and federated e-infrastructures. ● Thematic infrastructures. Thematic infrastructures create a shared and collaborative research environment, known as the RI ecosystem, which has shaped big science for decades. Examples include the European Organisation for Nuclear Research (CERN), the European Southern Observatory (ESO), etc. RIs manage a large amount of data and have often triggered the development of data technologies and related policies. ● RI clusters. RI clusters are groups of RIs horizontally interlinked to be able to address globally important scientific and technological challenges. They have strong links with research communities and projects, manage significant data volumes and develop innovative data analytics tools, ensuring effective research data exploitation. Five - 54 - ESFRI cluster projects have been launched in 2020, providing a focus for various ESFRI projects and landmarks to connect to EOSC. In general, the expectation of EOSC raised in the position papers of ESCAPE, PaNOSC, ENVRI-FAIR, EOSC-Life and SSHOC is that EOSC would enable the accessibility and reuse of research data and increase its scientific value. The landscape-surveying exercise continued until the end of 2020. In its entirety, the activity indicated which infrastructures were to be considered to be the key elements of the future shape of EOSC. The differences among particular European states should be taken into account. This time-demanding process has made EOSC implementation gradual and dynamic. Links between the national thematic infrastructures and e-infrastructures, including data infrastructures, and national open access (OA) repositories, have been investigated. The readiness of the states depends upon acceptance of EOSC. 2.6. International dimension EOSC operates in a global ecosystem with the clear aim, as already described above, to promote the ‘Open Science, Open Innovation and Open to the World’ principle in its international activities. The international dimension of EOSC is framed by the (i) regulatory framework, the Acquis Communautaire, (ii) Open Science culture, as well as (iii) the existing infrastructures and initiatives of the Economic Partnership Agreement (EPA) members. Open Science is a new era in the evolution of science, which requires a cultural shift. It is driven by a number of organisations, both long-standing, who are in the process of adapting their methods to the new developments, and recently set up, via a bottom-up process. The transition and expansion of Open Science presents a constant increase in scale and scope for science at the local and global level, resulting in a peak in the need for investment, limited by public resources. The current level of integration in the field of science in the European Research Area allows EU Member States and Associated Countries to share the burden of investment to achieve Open Science in the ERA. Moreover, the investment of the EU in e-infrastructure in recent years allows the opening-up of EOSC to third countries, based on shared values, principles and conscious choice. Given the different approach taken in the regions, EOSC will need to offer a tailor-made approach, taking into account local capabilities and demands. The European Open Science Cloud is an opportunity to give fresh impetus to Science Diplomacy. EOSC does not exist in a vacuum. Regional and national Open Research Data Commons and/or Open Science Clouds are being established concurrently. These developments enable the EU to enhance scientific cooperation with other parts of the world and drive Open Science culture based on commonly agreed values. EOSC operates in a global system which influences the world and is influenced by parallel activities from around the globe. There are major global trends which can be observed and groups such as the RDA Global Open Research Commons provide a useful forum in which to identify these and exchange lessons learned. Many international organisations such as RDA, CODATA, WDS and GO FAIR enable tighter collaboration between global initiatives, working together towards common goals for Open Science, thus driving global convergence on standards. At the same time, regional Open Science initiatives are getting more aligned and coordinated and there is a willingness for collaboration to avoid the creation of Open Science silos. - 55 - That said, EOSC has been enshrining a number of principles regarding international cooperation, with which potential partners could and should comply. While these rules and principles may be seen as a burden or an exclusionary tactic, in reality these ground rules enable a competitive, transparent Open Science ecosystem that enables quality science. ● Data portability. EOSC will not allow vendor lock-in at the EOSC-Exchange level and expects the same from services provided by third country partners. ● Digital sovereignty. Participation of third-country entities in EOSC is on a voluntary basis, but if they do participate it is expected that they will comply with relevant legislation and rules. ● Ethics and values. EOSC and the European research community represent certain values. EOSC recognises, however, that these might differ from those of other countries and is open to investigate whether its ethics and values should be reassessed in the context of globalisation. ● Individual and community data autonomy. EOSC condemns digital feudalism and supports the Global Indigenous Data Alliance (GIDA) and the CARE principles [GIDA; CARE]. ● Interoperability. The EOSC-Core will provide an infrastructure with basic functionalities, such as persistent identifiers (PIDs) or authentication and authorisation infrastructure (AAI) services. Research outputs will have to comply with the FAIR principles and services will have to be FAIR-enabling. EOSC will use open source solutions but will of course make some technology choices regarding the fundamental functionalities. Third-country participants who wish to participate as a user or service provider will have to comply with these requirements. ● Reciprocity. Reciprocity is a principle enshrined in the future International Cooperation rules of the framework programme. International partners to EOSC should provide access to their National Open Science Cloud or similar, and to their service portfolio. This would enhance the free flow of (research) data and services. ● Security. Third-country participants accept the cyber-security levels set by EOSC and commit to a Code of Conduct in the EOSC ecosystem. ● Openness. Third-country participants participate voluntarily in EOSC as users or service providers. 2.7. Strengthening the community 2.7.1. EOSC Association and EOSC Partnership The decision towards a Co-programmed European Partnership on EOSC required the incorporation of a new legal entity able to enter into a contractual arrangement / Memorandum of Understanding (MoU) with the European Commission: the EOSC Association. The EOSC Association was established on 29 July 2020 as a not-for-profit international association (AISBL) in Belgium, involving research and innovation stakeholders across the EU and beyond. The Association is open for membership and observership to organisations that embrace the vision and values of the Association and have a substantial and significant interest in, as well as can potentially contribute to and have an impact on, EOSC. - 56 - The EOSC Association plays a crucial role in gathering EOSC stakeholders such as research funders, policy makers, research-performing organisations and operators of research infrastructures to contribute to and monitor the future EOSC developments. The Association provides a single European voice for the purpose of advocacy and representation of all stakeholders in a collective manner. It facilitates communication, outreach and engagement with its members and observers, external service providers, research communities, stakeholder organisations and society as a whole to assure transparency, and promotes Open Science for the benefit of all. The signature, by the EOSC Association and the European Commission, of the contractual arrangement / MoU marked the start of the Co-programmed European Partnership on EOSC under the Horizon Europe framework programme. This EOSC Partnership brings together all relevant stakeholders to co-design and deploy a European Research Data Commons where data are findable, accessible, interoperable and reusable (FAIR). An open and inclusive Partnership will help ensure directionality (common vision and objectives) and complementary commitments and contributions at all levels. It helps to provide a framework to reach consensus amongst those committed to achieving results. The EOSC Partnership aims to expand on the planned Minimum Viable EOSC (MVE) to create a growing ecosystem, bringing together relevant European initiatives around the FAIR data economy, fostering collaboration among those initiatives towards the objective of open research, attracting small and medium-sized enterprises (SMEs) and start-ups to use and benefit from the federated services and data sources, and raising awareness in society about the benefits of FAIR-data-driven innovation. The SRIA strategically defines the roadmap for implementing and further developing EOSC with the wider stakeholder community. The EOSC Partnership develops and implements EOSC through financial, in-kind and policy commitments. As a contractual document between the EOSC Association and the European Commission, the MoU sets out two tiers of commitments from its two signatories, reflected, for convenience, in this SRIA: 13 ● Tier 1 is the contribution from the European Commission. Tier 1 will mainly include both financial and policy commitments. Through its work programmes, Horizon Europe mainly launches open calls for proposals for the EOSC community to develop and implement EOSC. Horizon Europe also mandates and stimulates policies supporting EOSC and Open Science, such as publishing in open access journals and making data FAIR. The EOSC Association supports the Commission by providing input to guide the calls and policy instruments for EOSC, mainly through the SRIA versions and Roadmaps, in the context of the EOSC Partnership Board. ● Tier 2 is the contribution from the EOSC Association and its constituent entities, among which some are entities mandated by Member States or Associated Countries. Tier 2 mainly involves financial, in-kind and policy contributions from research-funding 13 The SRIA Version 1.0 was initiated by the EOSC Executive Board under Horizon 2020 and has been developed collectively with the EOSC Community. This SRIA Version 1.1 is the result of integrating the MAR 2023-2024, ratified by the EOSC Community, into SRI v1.0. During 2023, SRIA Version 2.0 will be developed, in concertation with the EOSC Community and the European Commission. - 57 - organisations, research-performing organisations, service-providing organisations, as well as other organisations that are members of the EOSC Association. In addition, the EOSC Association makes a best effort to ensure contributions from other stakeholders in the EOSC community. These can be: ● Member States, Associated Countries and their respective national funding organisations which will directly finance research infrastructure developments in their respective countries that are compliant with EOSC standards and are easily integrated into EOSC. These financial contributions in the country contribute to building and growing the EOSC ecosystem. Member States, Associated Countries and national funders also develop policies that stimulate and support existing organisations in their countries to be as compliant as possible with EOSC and work towards sharing their research assets through EOSC. This includes policy changes that allow services to be made available across national borders, as well as aligning and supporting requirements for researchers such as data management plans, metadata standards, and making data FAIR. Such costs should be specifically earmarked for EOSC activities rather than general research activities in the country. ● Other individual organisations providing in-kind and policy contributions. Researchperforming organisations ensure that their data is made FAIR and linked to EOSC via trusted repositories that are compliant with EOSC standards. They also train and support researchers in Open Science by, for example, hiring professional data stewards and mandating data management plans and FAIR data. They further promote Open Science practices and the use of EOSC among their researchers, as well as implement new systems that recognise and reward researchers for doing Open Science in career and grant evaluations. Service-providing organisations integrate their services with the EOSC-Core, abiding by the Rules of Participation for EOSC. This increases the value-added service offer of EOSC and ensures composability of services to fully address the needs of researchers. The Association supports these organisations by sharing current developments, policy recommendations and best practices. One of the primary tasks of the Association is to continuously develop the SRIA, which shall influence future EOSC activities at institutional, national and EU level (including the EOSCrelated work programmes in Horizon Europe). The SRIA is not a static document, but rather is envisioned to be a living document that will adapt to the changing EOSC ecosystem and the needs of EOSC stakeholders in future versions that will be updated by the EOSC Association with input from the European Commission. The Association supports EOSC’s mission of enabling seamless access to data through interoperable services that address the entire research data lifecycle in a number of ways. It identifies key infrastructure requirements for the representation, capture, storage, processing and appropriate sharing of diverse forms of data by engaging with stakeholders and service providers. It enables key services, including but not limited to e-infrastructures, to promote broad and secure access to data resources and data processing services, through its role in shaping the relevant parts of the Horizon Europe work programmes and in monitoring the output of funded actions. It coordinates and fosters technical environments and promotes the skills that enable the federation of existing and new scientific data infrastructures. The Association’s Partnership with the European Commission and engagement with its - 64 - reusing the digital object the machine needs to know what it is allowed to do with it (‘know which actions are allowed’). This all needs to be well described in the metadata. FAIR is the set of requirements that ensures that digital artefacts within EOSC can be discovered and reused. The FAIR principles articulate a set of mutual responsibilities between content creators and curators. Digital artefacts must be described with rich metadata, assigned a globally unique persistent identifier, and be released with a clear and accessible usage licence. There is an onus on researchers to adopt relevant community standards and select appropriate data services that enable digital artefacts to be discovered and retrieved using standard protocols, applicable for both humans and machines. Research communities need to define standards, sharing agreements and services to enable FAIR digital objects. Some, such as astronomy, life sciences and linguistics, have self-organised, but many others require support in order to narrow the gap between communities. The strength of the FAIR principles is in defining a set of common characteristics required for all digital artefacts, irrespective of type, discipline and content. This enables machines to act across a broad set of content, enabling interdisciplinary research. Many aspects of the FAIR principles, however, address community-specific standards and practices. The principles will be applied differently according to the needs and requirements in the different fields of knowledge. Crosswalks and brokering are needed to support interoperability across the standards of multiple disciplines, as noted in Recommendation 4 from the FAIR Expert Group: ‘Develop interoperability frameworks for FAIR sharing within disciplines and for interdisciplinary research’ [EC_EG_FAIR]. Research communities need to be encouraged to develop and maintain interoperability frameworks that define their practices for archiving, referencing and describing research artefacts of all forms. To support interdisciplinary research, these interoperability frameworks should be articulated in common ways and adopt global standards where relevant. Intelligent crosswalks, brokering mechanisms and semantic and other technologies such as artificial intelligence, should all be explored to break down silos and allow cross-disciplinary exploration, analysis and visualisation. 3.4. Establish a sustainable and federated infrastructure enabling open sharing of scientific results This objective aims to enable the core functions of an operational EOSC ecosystem. EOSC is envisaged as a federation of infrastructures, forming a Web of FAIR Digital Objects and Related Services for Science. The FAIR principles and metadata standards act as guidelines for interoperability and facilitate maximum sharing and exploitation of research by the academic, private and public sector. The system will be based on three layers: (1) the federating core (or EOSC-Core), (2) the federation of existing and planned research data infrastructures, and (3) a service layer comprising common services and thematic services (EOSC-Exchange). Building on existing research data infrastructures, EOSC will grow through a series of iterations. Each iteration will add more functionalities and services for a wider user base and satisfy a broader range of use cases. (1) The EOSC-Core assembles all the basic elements to operate and provide the means to discover, share, access and reuse data and services in a reliable manner. These elements address key technical, cultural and policy decisions of EOSC and they must be maintained over the long term. Specifically: - 65 - ● A mechanism for naming and locating documents, data, software and services; ● A mechanism for discovery of and access to documents, data, software and services; ● A common framework for managing user identity and access. The EOSC-Core will need to assemble a number of basic services and features, including: ● Repositories complying with an open charter that describe what users can expect from the service, such as descriptions of the content with rich, community-defined and FAIR metadata (including granularity levels, versioning policy), sustainability commitments, quality goals, etc.; ● Networking connectivity with commitments on upload and download capabilities; ● Authentication and authorisation rules and services for allowing access by users. These rules and services have to comply with the EOSC authentication and authorisation infrastructure (AAI) standards; ● Persistent identifiers (PID) services complying with the EOSC PID policy; ● Metadata services describing the content available in order, for example, to allow discovery by end users; ● Application programming interfaces (APIs) for access by machines. These APIs are necessary to allow the development of applications using the content. Their description must be public. (2) The FAIR principles and metadata standards enable the federation of existing and planned research data infrastructures, adding a soft overlay to connect them and forming a Web of FAIR Data and Services. As the national, European and international research data infrastructures composing EOSC and other regional infrastructures are by definition distributed, as well as supported by a wide variety of institutions (public and private) throughout the world, the envisioned EOSC can only be realised in a decentralised federated way. As described above, this requires an underlying framework based on commonly agreed, minimum standards and maximum freedom to operate with agility, whilst still ensuring global and interdisciplinary interoperability. This does not rule out multiple ‘portals’ in the sense of more traditional websites, where users can enter the EOSC environment, find content and related services, learn about commonly adopted approaches, formats, standards and EOSC Rules of Participation, register their resources, tools and services, etc. Currently, the projects funded under EOSC-related calls in Horizon 2020 have developed an initial EOSC Portal. In order to enable innovative value-adding services to be developed, it is essential that such access points have an API for machine access. (3) The EOSC-Exchange builds on the EOSC-Core to ensure that a rich set of services (common and thematic), exploiting FAIR data and encouraging its reuse, are available to publicly funded researchers. It is expected that rivalrous services, such as those that store, preserve or transport research data as well as those that compute against it, will be made available via the EOSC-Exchange. Service providers that participate in the EOSC-Exchange will be required to conform to predefined Rules of Participation. ● Common services. This layer is composed of services that need to exist but may not be shared by all stakeholders. The main reason for such a layer is that certain domains or countries have already developed those services. There is no reason for them to change, while other domains or countries would benefit from using common services rather than developing their own. A good example is the archival service. All domains - 66 - and countries need archival services to ensure the sustainability of their artefacts (publications, data and software). Some stakeholders have developed their own and have no reason to change. Their experiences may, however, be useful in developing common services for other stakeholders. ● Thematic services. This layer has no limit. It covers all the services that communities need to develop to contribute to the EOSC ecosystem. These services are delivered to researchers and all stakeholders to enhance their working environment. They are built using the relevant elements of the federating core (EOSC-Core) and may leverage common services. They will use the APIs mentioned above when necessary. Many projects are already engaged in such developments in vertical domains. The mission of EOSC is to allow those services to flourish and to support the ecosystem while stimulating the creation of new innovative services. This section has explained the three overarching objectives that are driving EOSC and that will alleviate the challenges involved in delivering the European deployment of Open Science. The following section presents the guiding principles that are shaping EOSC and that will help position it within the Horizon Europe programme. - 67 - 4 Guiding principles 4.1. Introduction If EOSC is seen as the European endeavour of sharing research data, then this complements the European means to handle these data: the e-infrastructures in Europe. Data without e-infrastructures to store, compute and connect are of no use to EOSC and can only exist on paper or in the researcher’s head. On the other hand, e-infrastructures without any data (only ‘zeros’ or ‘ones’) are meaningless. Dataand e-infrastructures form what can be thought of as a ‘Yin-Yang’ relationship. One is not possible without the other. Whether the whole of data-infrastructures and einfrastructures should be called EOSC or whether EOSC is only a part of that is largely a question of semantics. The overarching principle for developing EOSC is that research has to be at the centre of the EOSC initiative. Thus engagement with research communities is fundamental to understand their requirements and ensure that the way in which EOSC operates and the services are offered is of use and value to the researchers and respects the academic sovereignty of research data. Communities need to be consulted and encouraged to take an active role in the EOSC ecosystem. They need to represent the diversity of practice, such as research infrastructures, universities, data stewards, research software engineers, professional associations, research leaders and early-career researcher organisations. Close attention needs to be paid to the existing standards, infrastructure and support within research communities and EOSC will recognise and adapt, where possible, to enable these. The diversity in readiness levels to adopt FAIR and Open Science principles will be taken into account, including assisting those communities that are less advanced. A good relationship with research-focused stakeholders needs to be ensured. Researchers have to be present in the various EOSC governing bodies to ensure acceptance by research communities and assist in promotion and advocacy in relevant fora. Given the diversity of practice and readiness levels, there is a need to ensure this is understood, accepted and represented within EOSC. It may be hard to engage researchers directly, so EOSC often has to work through intermediaries who can represent their requirements and interests. Within the overarching principle of the centrality of research and researchers, the way in which EOSC proposes to operate is defined by five further guiding principles. Over the last five years, as EOSC was in the making, a number of shared principles have emerged from the work accomplished by the European Open Science community. High-Level Expert Group reports and results from first-generation pioneering projects have fuelled the debate among the EOSC community. From this debate, a set of five guiding principles has been agreed upon which will help position EOSC within the Horizon Europe programme during the next seven years. These are: ● Multi-stakeholderism – EOSC will succeed if and only if it follows a multi-stakeholder approach; - 68 - ● Openness – EOSC will ensure that research artefacts are ‘as open as possible, as closed as necessary’ 14 ; ● FAIR principles – EOSC will assemble research artefacts that are findable, accessible, interoperable and reusable; ● Federation of infrastructures – EOSC will federate existing and upcoming research infrastructures (dataand e-infrastructures); ● Machine-actionable – EOSC will strike the right balance between machines and people in delivering the services that will serve the needs of European scientists. The following sections report on: ● The roles played by the wide range of stakeholders of science (Section 4.2); ● The two key ingredients of Open Science: ‘openness’ and ‘FAIRness’ (Sections 4.3 and 4.4); ● The way to federate the efforts of research, dataand e-infrastructures to serve the needs of scientists (Section 4.5). After recognising the role that machines will play at the service of scientists (Section 4.6), the final section (4.7) concludes with recommendations for research communities and policy makers on how to move towards the full implementation of Open Science, both culturally and technologically. Those recommendations can be seen as setting high-level requirements for the action areas that are presented in Sections 5 and 6. 4.2. Multi-stakeholderism Today, all scientific communities generate growing numbers of research digital objects of all kinds, from raw data to publications, including workflows and software. Over the last decade or so, there have been significant investments across Europe in computer-oriented research infrastructures and e-infrastructures. The outcome is a vast quantity of infrastructure components of various scales and scopes, centralised or distributed, generic or domainspecific. The challenge for EOSC is to federate this large variety of platforms at the subdomain, domain and interdisciplinary levels and to deliver an inclusive virtual environment to the European researchers. Most of these components have not been initially designed to work together. The challenge is not limited to linking datasets, federating infrastructures or aligning policies. It starts by linking multiple stakeholders – people and organisations – throughout the data lifecycle and across the EOSC ecosystem. At the same time, EOSC intends to address common political priorities of the European Union and its Member States, to: ● Make Europe fit for the digital age; ● Interlink data spaces across a more efficient European Research Area; ● Make Open Science mainstream in the research community. FAIR-by-design research outputs, combined with top-class digital infrastructures and artificial intelligence solutions, will ensure a true European research capacity to tackle the Sustainable Development Goals (SDGs), to reach the EU’s ambition for the Green Deal and to implement other national or sectoral policies. Furthermore, its domain-agnostic objectives to federate 14 For research artefacts, the interpretation of “as closed as necessary” is “as restricted as necessary”. - 69 - infrastructures and develop a Web of FAIR Data and Related Services bring new potential to contribute to the Horizon Europe missions and clusters. However, ensuring impact on these policy targets requires engaging further with a wide diversity and large number of stakeholders, across borders and disciplines, who are involved in the generation, storage, curation and processing of research artefacts, as well as in research policies, funding, skills and education. The EOSC Partnership aims to embrace such a multi-stakeholder approach, to provide a framework of collaboration and to pool and align resources at European, national, regional and institutional levels. Starting from the current Strategic Research and Innovation Agenda (SRIA) document, a central task for this Partnership will be to develop, update and monitor a holistic SRIA supporting the EOSC vision. Developing and implementing such a SRIA requires the involvement of a wide range of stakeholders, including: ● Member States; ● Research-performing organisations; ● Research infrastructures and e-infrastructures (e.g. related to storage, computing and communications); ● Research libraries; ● Research associations; ● Research-funding organisations; ● Etc. A more complete description of the EOSC stakeholders is provided by the EOSC Landscape report [EOSC_Landscape]. The approach to the implementation of EOSC during its initial phase has largely focused on EU-level activities 15 carried out through Horizon 2020 (H2020) projects. The consortia involved have brought together institutions of all sorts, from all over Europe and beyond, that have been developing and testing solutions along the six action lines described in the EOSC roadmap (data, services, architecture, access, rules and governance) [EOSC_Roadmap]. This EU grant-based approach has been successful in involving hundreds of European stakeholders across borders and communities, and has confirmed that the EOSC mission cannot be accomplished in a centralised manner but rather with a multi-stakeholder approach. In order to bring forward an initial operational EOSC capacity during the next phase of implementation, the EOSC Partnership will ensure directionality (common vision and objectives) and additionality (complementary commitments and contributions at EU, national and institutional levels). The willingness of the EU Member States and Associated Countries to embrace this coordinated multi-stakeholder approach was expressed in December 2019 by the EOSC Governance Board and confirmed throughout the development of the proposal for an EOSC Partnership. This reflects a broad interest on the part of the EU Member States and Associated Countries in making EOSC evolve from a call-based approach to an all-encompassing ecosystem where the different stakeholders make the necessary commitments to contribute, and deliver outcomes, on the most suitable level of intervention (EU, national, institutional). Doing so, this multi-stakeholder approach should strengthen ownership by the research 15 Total EU investment of about €320 million in the period 2017–2020. - 70 - communities, achieve scale by aggregating demand from researchers and other users, and pool existing capacities and expertise at all levels. All relevant research and innovation stakeholders, including scientific communities, research institutions, learned societies, community fora, national and international infrastructures (generic or thematic), funders (public or private) and industry actors (including data, software and journal publishers) are ultimately welcome to join the EOSC Partnership, if they agree to the Rules of Participation and adhere to the guiding principles, and to take part in the development of the present strategy and have their voice heard. In summary, developing wider synergies between multiple EOSC stakeholders and ensuring systematic and structural collaboration between the EOSC stakeholders will be essential to realise the EOSC ambition. This has resulted in the proposal to create an EOSC Partnership [EOSC_PP] and is reflected in the strategy put forward by this SRIA, in which there is not a ‘one solution fits all’ to address all the gaps and priorities to achieve the EOSC objectives, but rather a coherent compendium of activities and deliverables that will take into account the following: ● The most suitable level of intervention (EU, national, institutional); ● The main targeted categories of actors (research-performing organisations (RPOs), research-funding organisations (RFOs), service providers, policy makers, regulatory agencies, research infrastructure (RI) operators, e-infrastructures, libraries, industry, etc.); ● The most suitable programme(s) (Horizon Europe, Digital Europe, Connecting Europe Facility (CEF), structural funds, recovery plan, plus non-EU programmes); ● The most adapted set of instruments, also considering the full range of research data infrastructure (RDI) activities from academic research to innovation (coordination, research projects, demonstration projects, strategic workshops, etc.); ● The range of outputs: new knowledge, prototype solutions, guidelines, standards, services, infrastructure, training material, curricula, coordination, etc.; ● The expected impacts, including support of the Commission priorities and relevant EU policies. 4.3. Openness: ‘as open as possible, as closed as necessary’ 16 The rise of the digital age allows the ways research is conducted to change in multiple directions, with three main benefits: ● Delivery of better research results; ● Improved trust in research results; ● Development of multi-disciplinary programmes to address new societal and global challenges. However, these improvements will only materialise if scientists evolve their practices and look ahead to share knowledge in ways that take advantage of the new capabilities offered by the digital revolution. At the core of these practices is openness. Scientists need to embrace the new approach, where knowledge is shared at all stages of the research lifecycle, as opposed to the old way, where results are shared primarily through publications made available when the work has achieved a sufficient maturity level. 16 For research artefacts, the interpretation of “as closed as necessary” is “as restricted as necessary”. - 71 - 4.3.1. Open access The open access movement was born when scientists started to use digital technologies to share publications when they were still in preprint form (i.e. ready to be shared but not yet peer-reviewed). The time has come when this initial step can be followed by sharing not only publications but also all other research outputs such as data, software, workflows, etc. Open access has been fully endorsed by the European Commission in FP7 and H2020 programmes, first regarding publications and then extending the mandate to research data, and it is set to stay as a best practice in knowledge sharing and communication. Since 2012, when the EC Recommendation on access to and preservation of scientific information became available [EC_Rec_C(2012)4890], many Member States started discussing the need for establishing guidance and mandates on open access at the national level. In addition, institutions developed their own open access policies. It is now very common, for example, for research funders to require open access to research outputs for funded projects, to monitor research impact and the return of investments in research. Similarly, the Plan S initiative was launched in 2018 by a group of research funders [Plan_S]. It requires that scientific publications resulting from research funded by public grants must be published in compliant open access journals or platforms by 2021. This initiative is boosting the activities around open access publishing, in addition to fostering the discussion around roadmaps to establish open access and open science national plans and strategies in most EU Member States. Despite the momentum behind open access, though, a monitoring mechanism to check policies’ alignment and compliance with EC directives is still missing. Moreover, guidance is needed on issues such as security, privacy, property and sovereignty to ensure compliance between national and EC directives. The coronavirus pandemic showed, even more emphatically, the need for appropriate licensing practices to mitigate exclusive rights in copyright law [LIBER_Copyright]. Last but not least, some cultural and technological barriers still exist. On the researchers’ side, a strong bias still exists around open access publishing, which is often considered not comparable to traditional publishing. At the technological level, systems interoperability, enabling metadata exchange and improving dissemination and accessibility of research outputs, has improved recently, but there is still the need for further efforts to make the open access ecosystem a reality. While it is clear for most stakeholders that Open Science practices will improve research results (by allowing scientists to benefit from each other’s ongoing efforts), the work needed to realise the two other benefits – trust in science and multi-disciplinary developments – is still underestimated. 4.3.2. Trust in science through improved reproducibility As the world has become more complex, as human knowledge has expanded in more and more disciplines, the role of science has increased while becoming more and more difficult to follow for any individual. It is therefore essential for everybody to be able to trust research results in order for science to deliver its benefits for society. - 72 - In order to build that trust in science, research has to be reproducible. Reproducibility is the ability for an experiment or calculation to be duplicated by other researchers working independently. The reproducibility of science has been recognised as essential since the seventeenth century and the emergence of the scientific method. However, in the digital age, achieving reproducibility has become more difficult since computers have become part of the research lifecycle. Reproducibility of science requires reproducibility of software, as well as the availability of data and any other relevant information in machine-understandable form. Achieving reproducibility of science in the twenty-first century requires openness of software in all dimensions, not only source code but also knowledge of the computing environment. Use Case: The interdisciplinary European coastline expedition (TREC) as a pilot for managing multi-modal data across institutions, disciplines and countries Marine ecosystems are fundamental to life on our planet. Coastal ecosystems are particularly rich in diversity and abundance of both pelagic and benthic organisms. However, human activities are endangering coastal ecosystems on a massive scale. Climate change, in combination with habitat destruction, chemical and organic pollution along the coast and through river intake, and overfishing, represent major threats to coastal ecosystems. In order to characterise marine and terrestrial ecosystems along the European coastline, a European coastline expedition (TREC) is envisaged to start in 2022. It will also extend to freshwater as well as important littoral and terrestrial coastal ecosystems such as tidal flats, rocky coasts and marshland. Organised as a highly cooperative and large-scale multinational initiative, several key European research infrastructures such as the European Marine Biology Resource Centre [EMBRC], the TARA Ocean Foundation [TARA], with its research vessel, as well as national infrastructures such as the French Ifremer [Ifremer] will be involved as partners in this project. TREC will also cooperate with eLTER [eLTER] and multiple local partners who are invited to join the expedition through co-funded plug-in projects. The TREC expedition will sample multi-modal datasets along the entire European coastline, involving and combining multiple kinds of -omics and imaging data as well as environmental metadata. The multi-disciplinary nature of the project, representing different scientific communities, research landscapes and cultures, will require huge data integration and storage efforts. Extended longitudinal sampling by local cooperation partners will produce expanding datasets over the years. The storage and management of multi-modal data across projects, laboratories, institutes, scientific disciplines and countries will be a huge challenge to the project. The sharing, joint analysis and diverse reuse of acquired data are main pillars of the TREC research strategy but will need adequate IT infrastructure. This project responds to the challenges and threats that climate change in combination with habitat destruction, pollution, and overfishing pose to coastal ecosystems. A use case such as the TREC project will demonstrate the relevance of EOSC. In such a multi-disciplinary project of societal relevance, success will crucially depend on efficient sharing of data and processing pipelines at large scale, data standardisation, data access across domains, storage and service facilities, and accessibility to involved researchers and institutions, as well as other scientists. Access to, linking and integration of data and metadata from different disciplines on the same ecosystem should be facilitated and accelerated to ensure interoperability of national and - 73 - regional efforts. EOSC could be of help here, e.g. in the long-term cataloguing of data resources within EOSC. The expected outcomes of this use case will also be of value for activities around the European Green Deal, as cross-linking this data with socio-economic responses to changes in that ecosystem and other social science and humanities data promote an integrated understanding of the impact climate change and anthropogenic interventions have on European coastal ecosystems and the people living in and off them. 4.3.3. Facing global challenges through multi-disciplinary programmes In a world that has become instrumented, interconnected and intelligent, it is possible to launch multi-disciplinary initiatives where scientists from different domains collaborate. In order to benefit from research artefacts coming from different disciplines, machines are used to allow computations optimised with diverse sources. It is therefore essential that research artefacts are both open and machine-understandable. This requires not only the sharing of data and software but also the sharing of metadata that describe the research artefacts. Openness needs to extend to ‘information about information’. In many disciplines, efforts have been applied to design and archive ontologies that are becoming standards. In order to conduct multi-disciplinary projects, it becomes critical to develop crosswalks between metadata standards that will allow the matching of data representation designed for different domains. Openness of crosswalks themselves is therefore necessary to conduct multidisciplinary initiatives and is a key concept in the EOSC Interoperability Framework. 4.4. FAIR guiding principles: making science transparent and reproducible The FAIR principles were born with research data. Today, applying FAIR principles has to be extended to the whole research lifecycle, to ensure transparency, assessment, attribution and reproducibility. For this to happen, all outcomes of science, such as data, software and other digital outputs, have to be FAIR. 4.4.1. Web of FAIR Data and Related Services for science EOSC is conceived as a Web of FAIR Data and Related Services for science. This is intended to highlight the interconnectedness of people, services and content. For research data to have context and meaning, its provenance, quality and usage need to be shared. Who created the data? For what purpose? How has it been processed? Can it be trusted? Detailed metadata are required to enable discovery and reuse. The term ‘Web of FAIR Data’ is applied in its broadest sense, not just to data, but also to code, publications and other digital outputs. Services and stakeholders also need to be identifiable and well-described, with open metadata and persistent identifiers (PIDs) to allow cross-linking. The FAIR ecosystem proposed in the ‘Turning FAIR into reality’ report [EC_EG_FAIR] highlighted the importance of registries for various components, in particular for policies, data management plans, identifiers, standards and repositories. Sustainable funding for core infrastructure is required to support the principles of FAIR and openness. Sociological aspects also play a key role in the evolution of the culture and practices necessary to implement and benefit fully from the Web of FAIR Data and Related Services, in particular incentives and rewards to increase adoption of FAIR across communities, and the building of the necessary skills and specialised workforce. EOSC will be a federation of existing resources. It will of course give access to new data, but it will primarily be a federation of existing thematic data repositories and services, interfaced with existing data-sharing frameworks. To be adopted by the data providers and research - 80 - the forefront of establishing good practice in relation to data science, there have not always been clearly defined data policies in place to govern the generation, management and sharing of research data. Several of the EOSC Cluster projects 20 are working to define common data policies. In addition, there is an apparent need for a wider range of stakeholders across the research ecosystem to be involved in providing and maintaining this key information. Securing participation from different stakeholders will be vital to ensure that the profiles can be refined to better reflect some of the potential indicators emerging across the various EOSC groups, ESFRI, the Member States and the European Commission, and are incorporated into the evolution of the EOSC strategy in a timely manner. It is desirable that as the EOSC ecosystem matures, the content of policies is considered in addition to their existence. This will stipulate the evolution of national research environments, as the harmonisation of RI data policies is a valuable step towards supporting EOSC readiness. A few countries noted that efforts are under way to establish national-level research data competence centres. Such initiatives could play a significant role in coordinating EOSC preparations across RIs in a national context and potentially have a key role in monitoring ongoing levels of participation and performance against emerging indicators. Given that the landscape analysis indicates hundreds of infrastructure components available across Member States and Associated Countries that could potentially be federated, the EOSC Marketplace currently shows only a relatively small number of services per category. The reason for this is not clear. It may be the case that RIs that could provide services to EOSC have simply not yet completed the submission form required, and are waiting until they are sure of their readiness. While quality control measures are key for the longer-term delivery of the EOSC vision, it may mean that onboarding of services takes some time. A current offer of services and resources is managed by the EOSC Portal [EOSC_Portal] via a Catalogue and Marketplace [EOSC_Marketplace]. In that sense, the EOSC Portal serves as a possible future entry point to EOSC services and resources from many domains by enabling users to access and request e-infrastructures services and data supplied at institutional, national and regional levels, enabling them to process and analyse data in a distributed computing environment [EOSC_Svcs&Res]. In order to provide a rich platform offering a wide range of services and resources, the development of EOSC clearly requires the participation of service providers. Services and resources are provided and maintained by different providers under the observance of legal frameworks and under a variety of licences and access requirements [EOSC_Providers]. In spring 2020, the resources listed in the EOSC Portal Catalogue are offered by 73 service/resource providers and aggregators. Taken together, they would give access to: ● 254 services; ● 4.4M datasets; ● 141K software and applications; ● 34.6M publications; ● and 3M other research products. National and pan-European research infrastructures and RI clusters are quality and purpose assessed and horizontally interlinked to be able to address globally important scientific and technological challenges. They have strong links with research communities and projects, 20 [EOSC_Landscape] – section 3.5.1. - 81 - manage significant data volumes and develop innovative data analytics tools, ensuring effective research data exploitation. The potential scale and diversity of the services and resources implies that the operational and financial responsibility for federated services and data will remain with their existing funders and cannot be transferred to a central EOSC entity. The investment in federated services and resources by Member States needs to be measured and acknowledged as an inkind contribution to the overall EOSC funding model. EOSC can provide an environment driven by societal challenges for public and private sectors to co-design innovative data-rich services and, in turn, increase Europe’s technological sovereignty in key enabling technologies and infrastructures for the data economy. In a second iteration, the MVE can be expanded with additional functionality and services dedicated to the requirements of end users from the public sector, 21 who are not involved in research activities but want to exploit open access to research data. For example, EOSC can offer assistance to the public sector in relation to the implementation into national law by Member States of the Open Data Directive [EC_PSI] by July 2021. The scope of the Open Data Directive includes research data resulting from public funding and focuses on the economic aspects of the reuse of information. EOSC can also assist with the publishing of dynamic data, the uptake of application programming interfaces (APIs) and address the transparency requirements for public–private agreements involving public sector information, avoiding exclusive arrangements. The monitoring functions of EOSC could also help Member States identify high-quality datasets associated with important benefits for the society and economy. Enabling the private sector to make use of EOSC resources greatly increases the potential for innovation and economic impact of EOSC. Therefore, in a third iteration, the MVE can be expanded with additional functionality and services dedicated to the requirements of end users from the private sector, so that they can exploit the FAIR data and associated services for commercial gain without distorting market competition. 4.5.5. Future Outlook For EOSC to be a success, it must be widely adopted by researchers. This implies that EOSC must provide access to services that allow researchers to pursue their research activities more effectively through faster and seamless sharing of publications, data, software and other digital research outputs. While the services to be provided to researchers via EOSC are expected to be free at the point of use, 22 they are not without significant cost to build, maintain and operate. Researchers are practically minded and will only adopt EOSC if it provides interoperable services that make their research practices simpler and more effective, i.e. they need to be 21 In this document the term ‘public sector’ refers to all bodies governed by public law as defined in public procurement of services: Council Directive 92/50/EEC [EC_Procurement]. 22 Free at the point of use does not imply free of charge. Free at the point of use means the end user does not pay directly for the service when it is delivered, but their consumption will be paid for by other means. For example, an end user would not need to use a credit card to pay for a service but their employer may receive an annual bill from the service provider, or the employer may have arranged a suitable subscription. - 82 - easy to use and need to support all phases of the research lifecycle. Readily available training and documentation, employing the latest digital learning tools, will be needed to reduce the barriers to adoption. Therefore, the basic condition of success in ensuring EOSC sustainability is performance: how EOSC, as an ecosystem, operates and how the resources are provided, used and acknowledged by the users. Awareness needs to be raised among the EOSC stakeholder community of what is in EOSC and what is not, at all levels of the Interoperability Framework: technical, semantic, organisational and legal [EOSC_IF]. In order to gradually achieve interoperability of the services and compatibility of the data federated via the EOSC framework, standards and interfaces are needed, and the current activities and plans should put emphasis on developing those standards and interfaces step by step. This may involve revisiting and adjusting the datasets and e-infrastructures involved in the ongoing EOSC-related projects. Use Case: OpenBioMaps –a discipline-independent tool developed to foster relationships between science and applied fields OpenBioMaps (OBM) maintains an open and free biological database service and develops software applications for handling biological data. It offers scientists and conservationists a customisable toolset that facilitates easy access and management of data. By combining data from different fields and offering easy-to-use analysis tools, it widens uptake beyond academic research to applied fields and citizen science. In the majority of use cases, OpenBioMaps builds a connection between conservation and conservation biology. Data may be used to designate protected areas, track bird populations in cities, or to build road safety prediction models as a public service for some road sections. The data are also used in several citizen science and educational projects. OpenBioMaps has recently been enabled to use distributed computing resources from the EGI federated cloud through an EOSC service as part of the EOSC Early Adopter Programme [EOSC_EAP] to develop and maintain a service layer that allows OBM projects to run scientific analyses. As part of this programme, OBM is developing an interactive web interface that enables collaborative work on developing analyses. The project is also developing a portable computational package format (Docker container) which lets projects share analyses together with their environment so remote computational servers can run the analysis through these packages. The adoption of services available in the EOSC Marketplace will allow OBM to widen its service offering to users (researchers in conservation biology and ecology), enhancing the analysis capability and enabling greater cross-disciplinary applications of the data. - 83 - Additionally, standard ways of calculating costs should be created for services that also include margins and returns for service aggregators and other ‘intermediaries’ that are needed to de-risk the quality of EOSC services and cohesion between operators. Further consideration should be given to procurement processes (such as pre-commercial procurement and others) that could be used in order to eventually ensure that EOSC itself is able to buy all administration (and not only technical) services that it would need to survive. Procurement would be the mechanism determining the issue of intellectual property (IP) developed by EOSC projects. For projects funded through a grant mechanism with EU funding, the IP resides with the beneficiary who has generated the results (e.g. a university). However, it will be important that retention of results is with EOSC itself, to ensure its sustainability as well as the trust of the user community in the EOSC ecosystem. Finally, the success of EOSC depends not only on sound funding models encompassing the financial, legal and governance aspects to create added value for the stakeholders but, in accordance with preliminary feasibility investigations, also on the incentives and rewards for researchers that encourage them to participate in a culture of sharing the results of their research. Without such incentives and rewards it is possible that the uptake of EOSC could be jeopardised by lack of engagement from researchers. All of this will take time and cannot happen overnight. Implementing the set of iterations described in this document will take the EOSC schedule far beyond the end of 2020 and extend it over the full length of Horizon Europe, as depicted in Figure 4.2 below. Figure 4.2: Schematic representation of timelines of EOSC iterations Consequently, it is recommended that a transition period of 3 years (2021–2023) be anticipated to establish the MVE, building on projects to be funded via INFRAEOSC-03-2020 - 84 - and INFRAEOSC-07-2020 funding calls, and by means such as open calls related to the EOSC Partnership, other EOSC-related projects as well as contributions committed by the EOSC Association members. As of 2024–2025, EOSC will gradually open up to end users beyond the research community to develop and deploy services that will serve society at large, with significant contributions from the private sector. Specific activities and the timeline will be further elaborated taking into account the results of the previous period. It is expected that EOSC deployment will create market opportunities for new innovative companies to engage in the deployment of Open Science. The EOSC Partnership will also address the differences in economic development in the research and innovation sector by creating equitable access to data and services from both users and providers. Researchers and innovators will be able to jointly create innovative new technologies and services, which in turn will lead to the creation of new jobs and markets. The education, training and support needed to develop the necessary expertise will be facilitated by the use of virtual, shared environments. 4.6. Open Science services: machines in support of people The rise of the digital age creates new avenues for the development of Open Science, improving knowledge sharing between scientists. Digital technologies also allow new challenges related to the abundance of research outputs created around the world to be faced. Scientific activities have grown in volume and complexity in many ways. Machines are needed to help scientists face these new challenges. The volume of scientific results produced every day has grown significantly. Even within a single discipline, it has become impossible for any scientist to read all the publications related to her/his research. When it comes to multi-disciplinary research activities, the scope of knowledge is beyond reach for a single individual; teamwork is no longer an option. As the deployment of the internet extends the scope of research artefacts to publications, data and software, the volume of information available can no longer be managed by a research team. As an obvious consequence, research can no longer be done without the use of machinedriven systems (hardware and software). EOSC has to help scientists exploit those systems to perform their activities. Table 4.1 highlights the variety of systems that are commonly used and positions EOSC as offering an integrated view of those systems by federating existing infrastructures. - 85 - 4.6.1. Digital systems for Science Systems/Users Hardware Software Individual Scientist Personal Workstations, Tablets, Smartphones, Specific Devices, … Generic & Specific Applications Generic Software (e.g. operating systems, programming languages environments), … Research Team Computing & Storage Servers, Specific Equipment, … Databases, Shared Repositories, Shared Applications, Shared Libraries, … Research Organisation Large Computing & Storage Servers, Large Specific Equipment, … Development Platforms, Shared Repositories, … Research Infrastructure High Performance Computing, Very Large Storage, High Performance Equipment, … General Purpose Applications & Platforms, Very Large Repositories (publications, data, software) Table 4.1: Commonly used systems, of which EOSC offers an integrated view by federation 4.6.2. Hardware Digital hardware is managed like other research equipment. Sharing hardware has become common practice within laboratories, universities or research centres. Infrastructures allow resources to be shared at national or thematic levels. The deployment of EOSC requires sharing of resources across borders and across disciplines. Nowadays, thanks to the availability of internet-based infrastructures, the technical aspects of sharing resources can be addressed. The challenges to achieving the EOSC vision with regard to these hardware resources will mostly be at the legal, financial and organisational levels. The multi-stakeholder approach is essential to address these challenges. Agile agreements, shared funding models, deployed Rules of Participation are therefore foundational for EOSC to deliver its full potential. 4.6.3. Software In order to implement machine actionability, software is used at multiple levels of the research environment, which can be split into two categories: - 86 - ● Research software: software used by scientists themselves to manage experiments, collect data, exploit results, check hypotheses, etc. ● Infrastructure software: software used to manage infrastructures at the service of scientists. Each of these categories is described below, together with the software services on which the EOSC-Core itself is based. 4.6.3.1. Research software as research artefacts In order for digital systems to help deliver their value to scientists, research artefacts have to be machine-actionable. As described in Section 4.4, research data have to comply with FAIR principles in order for digital systems to be able to find, access and reuse those data. Sharing research publications also benefits from the deployment of digital services built on top of the World Wide Web, which was introduced originally as a Web of documents. While publications benefit from the Web of documents and the tools and practices that have been developed over the last thirty years, and research data benefit from special attention after the emergence of the FAIR principles, research software has started to receive attention only in the last few years. Research software does not benefit from similar opportunities for a variety of reasons: Research software has received recent attention It is common for scientists to evolve research software in order to conduct derivative research initiatives. Therefore, software, as for any other research artefacts, has to be archived, referenced and described in order to be reused. Reproducibility of science requires the availability of the exact software version that has been used by prior experiments. Publishers have started to include research software in their repositories. Data repositories have started to include software next to their datasets. From these pioneering efforts, a systematic approach to managing research software has to be put in place. During its transition phase, EOSC has recognised this situation. A task force on ‘Scholarly Infrastructures for Research Software’ has been organised by the Architecture Working Group and has delivered a report covering the state of the art, best practices and open issues, workflows and use cases before offering recommendations for next steps [WG_Arch_SIRS]. Research software benefits from generic software environments On the other hand, software benefits from the open source movement, which has been under way for decades. Open source allows software source code to be shared. As a consequence, one of the main characteristics of research software is that code reuse is considered normal practice. Moreover, thanks to the deployment of the internet, cooperative software development efforts are improved by the use of software development platforms. It has therefore become possible to harvest open source code and build open source software archives. Also, software development platforms use version control systems which allow the software evolution to be archived. Users can therefore retrieve the exact version that was used to produce the research results. Research software needs metadata description standards Research software is now recognised as playing a key role in research activities, as described by the CodeMeta project [CodeMeta]: - 87 - ‘Research relies heavily on scientific software, and a large and growing fraction of researchers are engaged in developing software as part of their own research. Despite this, infrastructure to support the preservation, discovery, reuse, and attribution of software lags substantially behind that of other research products such as journal articles and research data. This lag is driven not so much by a lack of technology as it is by a lack of unity: existing mechanisms to archive, document, index, share, discover, and cite software contributions are heterogeneous among both disciplines and archives and rarely meet best practices.’ The deployment of the World Wide Web has provided the opportunity for the development of general schemes to describe information as shown in Figure 4.3. Research software developments have created their own schemes. Figure 4.3: Software ontologies landscape derived from Pathways for Discovery of Free Software [Gruenpeter & Thornton] CC-by-4 After the creolisation period that has occurred recently, it is time to agree on metadata standards for software source code. Work is under way and EOSC will be able to both contribute to the standardisation and benefit from it. 4.6.3.2. Infrastructure software as service delivery Over the last decades, a wide variety of research infrastructures has used the availability of new delivery models to develop new-generation infrastructures for the benefit of scientists. Those infrastructures are organised at national and thematic levels. The Cloud Computing paradigm for sharing resources has developed and research infrastructures now have the choice when it comes to delivering their value to end users. Services can belong to one of the Cloud Computing layers as described in Figure 4.4. - 88 - Figure 4.4: Cloud Computing layers (from Wikipedia) When designing a new Cloud offering, the way resources are shared can belong to Cloud Computing types as shown in Figure 4.5. Figure 4.5: Cloud Computing types - 89 - There is no ‘one size fits all’ model for research infrastructures. Depending on the purpose of the infrastructure, the appropriate model will be chosen. As a consequence, EOSC will have to federate very different infrastructures. EOSC will be faced with the challenge of hiding the complexity and the diversity of services to the end user by providing a simple-to-use environment. Rules of Participation to the EOSC-Exchange are essential for resource providers to best deliver their value. 4.6.3.3. EOSC-Core federating Last but not least, while as lightweight as possible, the EOSC-Core itself will be based on software services. In order to deploy EOSC services in a controllable manner, special attention has to be given to managing the software involved in implementing EOSC-Core functionalities. The EOSC-Core needs to be exemplary in terms of openness at all levels: ● Open source code; ● Open interfaces; ● Open protocols; ● Open standards; ● … While a reference implementation is critical to bootstrap EOSC deployment, the evolution of EOSC should be driven by innovation practices and allow multiple implementations to be welcome. 4.7. Recommendations Starting from the guiding principles, it is possible to highlight recommendations for research communities and policy makers, to move them forward from the current state of the art towards an Open Science scholarly communication ecosystem that is based on, incentivises and facilitates Open Science principles and practices in performing and sharing science. Research communities should: ● Normalise their Open Science processes (standards); ● Regulate them (policies); ● Facilitate their implementation (guidelines and frameworks, e.g. information models that describe flows and elements); ● Make sure their thematic services embed Open Science aspects by design (roadmaps). The aim is twofold: to make the future scientific process ● As rigorous and automated as possible (e.g. services to FAIR-publish all outcomes on behalf of researchers); ● As transparent and reproducible as possible (e.g. tracking provenance, services, researchers, data, software, relationships, etc.). Scientific communities should share a common understanding of the research products they manage, how these are semantically related, and how these should be published in order to maximise their discovery, access and reuse. For example, the concept of ‘experiment’ should be published, with all the elements necessary to ensure its reuse, replicability, reproducibility and repeatability by others. - 96 - 5 Implementation challenges Based on the guiding principles and recommendations, the European Open Science Cloud governing bodies have identified fourteen action areas to help deploy the EOSC ecosystem. The seven areas relating to the primarily technical challenges and prerequisites to implementing the EOSC ecosystem are: ● Identifiers; ● Metadata and ontologies; ● FAIR metrics and certification; ● Authentication and authorisation infrastructure; ● User environments; ● Resource provider environments; ● EOSC Interoperability Framework. This section describes each of those areas, giving an assessment of status, identifying gaps and proposing priorities. The remaining action areas are described in the next section, Boundary conditions. 5.1. Identifiers The persistence of the identity of digital objects and stability of references to those objects are essential to the European Open Science Cloud. Only if researchers can be assured that digital objects (including publications, data and software resources) do not alter over time and are continuously accessible via linking mechanisms can a trusted distributed research ecosystem that supports verifiable and reusable research be sustained. The use of persistent identifiers (PIDs) has been specifically recognised within the FAIR principles as a key feature supporting the findability and accessibility of research objects. PIDs therefore form a stable, trusted structure which can be used to make the research infrastructure a reliable source of verifiable and reproducible research. EOSC should seek to support a shared policy for the use of PIDs both for the management and analysis of data, and also for the publication, curation and tracking of research outputs. 5.1.1. Status Systems that are based on an uncontrolled assignment of identifiers prove to be too unstable for trustworthy long-term identity management. In order to provide trusted PIDs that are usable, a combination of organisational and technical solutions needs to be supported. Services need to supply PIDs that are globally unique and have a stability of reference over time, and thus require organisational management and ongoing support. The EOSC PID policy [EOSC_PID_Policy] sets out the expectations on the use of PIDs and PID services by participants in EOSC. Persistent identity is an established field, and mature technologies (e.g. Handle [Handle]), infrastructures (e.g. DOI [DOI]) and organisations (e.g. DataCite, ORCID, DONA, ePIC [DataCite; ORCID; DONA; ePIC]) already exist to support PIDs. The issuing of PIDs for publications (e.g. Crossref PIDs [Crossref]) and their use within citations has become standard practice. The use of PIDs for data citation (e.g. DataCite) and also unique references to people (e.g. ORCID) has been extensively developed over the last decade and has become widely accepted practice in the research community, although their uptake and use by the research community at large is not universal. - 97 - Beyond publications, data resources and researchers, PIDs are of value to identify all resources used or referred to in research data and accompanying metadata. These could include documents, data, people, organisations, projects, funding, software, services, instruments, samples, videos and other artefacts. Using PIDs for these resources gives more reliable and semantically meaningful means to provide rich metadata to support research as well as properly attribute and track the use of valuable research objects. There are emerging technologies, standards and organisations to support many of these, although to date uptake has been limited due to a lack of commonly accepted approaches and clear business cases for their use. Data generation and analysis applications are also increasingly being required to access and process data on a large scale and across distributed infrastructure. Software tools need to address data objects reliably and PIDs provide a means to do this. In these applications, PIDs need to be issued and accessed rapidly and at scale at the data generation stage and to be accessed across the research lifecycle. PIDs thus need to be assigned at an appropriate granularity for the application, to support the addressing of data objects within a larger aggregation. Support for versioning and tracking through the data lifecycle would also need to be supported to accurately record the provenance of data from raw through fully qualityassured data to actual results. The concept of a FAIR Digital Object has been developed, with an inherent use of PIDs, and standard specifications of PID Kernel Information and PID Type Registries published [RDA_PID_Kernel; RDA_PID_Registry]. Local handle systems have been provided to support these use cases. Nevertheless, this is an area that needs further refinement both in the applications where it is of most value and the practical technologies involved. PIDs are thus an integral part of the research infrastructure, and play a key function in the data lifecycle, from the data generation and analysis stage to research output publication, curation and reuse. Tracking and connecting the use of PIDs in metadata and in citations, where they refer to one another, can form the basis of a rich, searchable resource for finding and contextualising resources. Tools that exploit this ‘graph of research entities’ include the Research Graph from OpenAIRE [OpenAIRE_RG] and the PID Graph from FREYA [FREYA_PG]. 5.1.2. Gaps PIDs are an established mechanism which has been used for nearly 20 years. However, there are still areas for further development. ● Establishing mature and recognised PID infrastructures for emerging resource types. There is a need to develop and establish trusted and widely used PID infrastructures for a wider range of resource types. In particular, instruments, software, organisations and services are types that would be of value in EOSC, although there is a wide range of further objects, such as physical infrastructure, physical samples, video recordings and theoretical concepts, some of which are domain specific. There are mature technologies for some of these resources, whilst others need development, and action on the adoption of all PID types is required. The PID types should then be used within core EOSC services, such as to register the scientific services in the EOSC-Exchange. ● Support for machine-actionable PIDs. Tools and standards supporting machineactionable PIDs have been developed over recent years, including PID Kernel Information and PID Type Registries, but are not as yet mature or widespread. PID Kernel Information has been introduced as a small amount of standard metadata - 98 - within the PID record to allow programmatic access and use [RDA_PID_Kernel]. PID Type Registries [RDA_PID_Registry] and accompanying Kernel profiles are not as yet standardised for different machine-readable data types and automated processing is largely missing or considered experimental. EOSC should consider the support of PID Type Registry services within EOSC, and develop services and use cases that exploit these services in automatic data analysis. ● PID ‘meta resolver’. Each PID provider provides its own resolver, while a meta resolver could form a single service which can recognise different PID types and redirect to the appropriate resolver, regardless of issuer. ● Standardising the PID graph. Tools for connecting and searching across networks of PIDs are still prototypical, with several different approaches being explored (e.g. Research Graph, PID Graph, CERIF [CERIF]), and services developed to exploit this PID graph are still experimental and local. There is a need to standardise approaches across PID providers and for the uptake of tools built on this graph to become more widespread. ● Integration of PIDs into FAIR data management. The use of PIDs should be integrated into workflows that collect and analyse data to ensure that FAIR data is generated. PIDs need to be assigned early and potentially at scale (depending on the application). Collection of metadata associated with a PID needs to be automated close to where the data is generated, and integrated into data collection and processing workflows. ● PIDs and sensitive data. The FAIR principles would require PIDs to be used with sensitive data and this would reflect onto the PIDs themselves. This may lead to situations where access to parts of the metadata is restricted. EOSC would apply the principle of ‘as open as possible, as closed as necessary’ 25 , and fine-grained access control for creating, updating and accessing PID records (Kernel information) may be needed. ● Quality of service for PIDs. The EOSC PID policy defines expectations on the quality of service of PID providers and services. The extent to which providers and services comply would need to be validated. The enforcement of policy is a governance rather than a technical issue for EOSC, but the governance may need to be supported by tools and processes to publish in a machine-readable form or validate the service. ● New PID technologies. New mechanisms and tools are appearing which support PIDs in novel ways. For example, some approaches do not require an authoritative certifying organisation. Intrinsic or smart PIDs are inferred (computed) from the form of the object (e.g. identifiers for software, or chemical objects). Others are decentralised, with no issuing authority but rather use distributed ledger technology to ensure their integrity. Further development and exploration should be encouraged within the EOSC programme. 5.1.3. Priorities ● Develop standardised identifiers for resource types that have not as yet become standard practice. For general research use, EOSC would prioritise identifiers for 25 For research artefacts, the interpretation of “as closed as necessary” is “as restricted as necessary”. - 99 - instruments, services, organisations and software, although there is a need for particular domains to provide their own community standards. ● Develop a ‘meta resolver’ that can deal with any type of relevant identifier. ● Define specifications (schemata) for PID records / kernel information to support machine-actionable PIDs. ● Produce type definitions for the most common data formats or building blocks. ● Provide standardised interfaces and protocols for exchanging information on PIDs to support the creation and use of a PID graph. ● Develop tools to support the certification of PID infrastructure against the EOSC PID policy. 5.2. Metadata and ontologies Metadata and ontologies are essential to realising Open Science, and thus are an important topic that needs to be addressed by EOSC. Metadata and ontologies have evolved organically over time, addressing the needs of individual communities and sub-communities. Because of these community-specific drivers, to date an overarching, coordinated approach to metadata and ontologies for scholarly resources has for the most part been missing. Interoperability is thus the biggest gap that EOSC needs to address with regard to metadata and ontologies. The EOSC Interoperability Framework is taking a broader approach to identifying gaps and setting priorities relating to interoperability, and this section on metadata and ontologies fully aligns with this broader approach. Without improvements in interoperability, there will be no widespread adoption of metadata schemas and ontologies in European research activities, and Europe will fall short of fully realising Open Science. The path towards better interoperability and adoption of existing metadata schemata is through the development of governance structures for how metadata and ontologies are used within EOSC. This governance should be built primarily around existing discipline-based communities but needs to be coordinated across these communities within EOSC, to drive the process of improved interoperability and increase adoption. Coordination with activities around metadata and ontologies outside of EOSC, for example in the Research Data Alliance (RDA), is of course essential. In addition, the interaction with the research information systems (CRIS) community and its metadata expertise will be of benefit in terms of enriching the research data with an elaborated set of additional research-related metadata (on researchers, projects, organisations, equipment, etc.) adding to the FAIR-ness of the data, while at the same time keeping the administrative burden for the research community as low as possible according to the ‘only once’ principle. The work that these governance structures coordinate should include registries that describe metadata schemata in a standardised and machine-actionable way, better researcherfocused tools and services working with these metadata, crosswalks between existing metadata schemata, and training and documentation. The drivers for all work regarding metadata and ontologies should be use cases from and adoption by the researcher community, and the work should be based on existing infrastructure and communities. 5.2.1. Status Scientific disciplines and communities have defined specific detailed metadata schemas and ontologies to describe community-owned data products. The adoption varies between - 100 - research disciplines and is, for example, strong in the life sciences (e.g. DICOM [DICOM]), or astronomy (e.g. FITS [IAU_FITS]). Metadata schemas describing resources that are not research outputs, e.g. organisations, instruments, samples, workflows or projects, have been developed in the research information community since the 1980s, resulting in standard metadata schemas that include projects, organisations and publications as well as persons (researchers), equipment and datasets, amongst others, while the schemas on samples, workflows or services are mostly an emerging activity. Integration of discipline-specific metadata across communities and the aggregation of metadata derived from different metadata schemas and ontologies is still lagging. Automatic metadata generation from instruments would be very beneficial, but is not yet common practice (though there is, for example, EXIF [EXIF], and again DICOM). 5.2.2. Gaps Work on developing, improving and applying metadata schemas and ontologies – both for specific disciplines and for general use – is happening in many different places, but is often not well-coordinated, leading to a number of standards that are sometimes not well-aligned or that even conflict with each other. Information about existing metadata schemas and ontologies is scattered across organisations and services, making it hard for users to find the relevant information. Such information is usually not described in a standardised way. The communities using a particular metadata schema are not always easy to identify. Communities have defined crosswalks to map different metadata schemas and ontologies, but there is no standard way to describe or discover these existing crosswalks, nor to facilitate their maintenance when updates to the schema are applied. Crosswalks between community-specific metadata and generic, common metadata, allowing the harmonisation of metadata for use cases such as discovery, have not been fully exploited, leading to silos of metadata that cannot be easily aligned. The development of solutions in which community-specific metadata and generic, common metadata are integrated, resulting in an optimal application of metadata for use cases such as discovery, needs attention. In this respect one could build on existing work on metadata models, in which generic, discovery and subject or field-specific metadata are combined and integrated. User-friendly tools to apply and maintain metadata for all types of research objects are not easy to find or are not available. 5.2.3. Priorities ● Develop governance structures for coordinating the work on metadata and ontologies within EOSC, both for specific disciplinary communities and for overall coordination. ● Provide or embrace/stimulate existing registries of metadata schemas and ontologies, defining clear protocols for federation/harvesting, crosswalks and tools for metadata management. - 101 - ● Engage with the research information community in order to maximise the re-uptake of the information already residing in research information systems that communicate in a semantically interoperable manner based on standards. ● Develop EOSC guidelines for a minimum metadata description based on existing metadata schemas and tools to allow data discovery and metadata exchange across federated repositories and scientific communities. ● Develop services that build on metadata registries and can facilitate the diffusion of metadata schemas across communities, sharing and community maintenance of crosswalks, measurement of metadata resources uptake across communities, validation of data sources against metadata schemas, etc. 5.3. FAIR metrics and certification 5.3.1. Status The FAIR principles are a recent concept so metrics are still under definition. The principles were intentionally articulated broadly but this ambiguity leads to different interpretations and the risk that metrics do not fit different community practice. The implementation of FAIR can only be achieved in an ecosystem. Research artefacts are made FAIR by the services in which they are created, discovered and reused. The FAIR principles therefore need to be applied to all components of the ecosystem, since FAIR data maturity depends on the capabilities and trustworthiness of services such as repositories and persistent identifier systems. The definition of criteria potentially has very significant consequences if they are used to decide on participation or funding. Also, metrics are not meant to be a punitive method for direct comparison between datasets from different areas, because communities will arrive at optimal FAIRness in different ways. These risks are well understood by the community: the open consultation on the SRIA held during the summer of 2020 showed that metrics and certification are given a low priority, ranking second-to-last with 39% of votes in the feedback compared with 78% for the highest-ranked priority, metadata and ontologies. This has to be taken into account, by implementing them inclusively and progressively, taking into account also that FAIR is a journey, the diversity of community FAIR practices and the highly different stages of preparedness of the communities, to enable buy-in by a diversity of communities. It is essential to examine the criteria applicability and to gather feedback in a wide range of contexts. The Metrics and Certification Task Force of the EOSC FAIR Working Group recommends that the definition of metrics should be a continuous process, regularly tested and iterated to minimise these risks. Inclusiveness should be a key attribute, to recognise the diversity of practice across communities and the different stages of FAIR maturity. Existing work, in particular by the international FAIR Data Maturity Model Working Group of the Research Data Alliance (RDA) [RDA_FAIR_DMMWG], should be built upon and tailored to the EOSC context. This forum also provides an appropriate international community to iterate and maintain the metrics, ensuring collective, community governance. Status of metrics The RDA FAIR Data Maturity Model Working Group has published a model with 41 criteria, allowing compliance of data with the FAIR principles to be assessed. A degree of priority – essential, important, useful – is attributed to each criterion. The Working Group has worked in a transparent way, and requested inputs and tests from the community throughout its eighteen-month time span. The model is being implemented, for instance, in FAIRsFAIR - 102 - [FAIRsFAIR], which is progressively defining criteria to deal with use cases. The Metrics and Certification Task Force of the EOSC FAIR WG is proposing a set of possible EOSC metrics as a target, with a timeline towards progressive implementation, which requires extensive testing by a wide range of communities. FAIRsFAIR produced a first assessment of FAIR semantics (semantics is discussed in Section 5.2 Metadata and ontologies) and high-level requirements for assessment frameworks, and an evaluation of how services influence data FAIRness. Software is another important component of the FAIR ecosystem. A Working Group, FAIR 4 Research Software [FAIR4RS_WG], common to the RDA, Force11 and the Research Software Alliance, was created mid-2020, as a result of discussions held in many venues during recent years. Its aim is to define the FAIR principles for research software and provide guidelines on how to apply them. This WG should bring another key component to the FAIR ecosystem. Status of certification As stated in the ‘Turning FAIR into reality’ action plan [EC_EG_FAIR], there is a need for certification schemas to assess all components of the FAIR ecosystem. Significant work has been devoted to certification of data repositories, with an international landscape that includes in particular CoreTrustSeal [CoreTrustSeal], which provides a generic core framework for trustworthy repositories and has now certified an international set of trustworthy repositories in different disciplines, DIN 31644 (nestor Seal) [DIN_31644; nestor_Seal] and ISO 16363:2013 (also known as CCSDS 625.0-M-1 – Audit and certification of trustworthy digital repositories) [ISO_16363]. In parallel, ELIXIR is developing its own evaluation badges and processes [ELIXIR]. The availability of certification criteria is also an asset enabling repositories to self-evaluate and improve their practices and processes, even if they do not apply for formal certification. In the context of FAIR, work is ongoing, in particular in the FAIRsFAIR project, on FAIR alignment of repository certification schemas. This is complementary to the evaluation of the FAIRness of the data itself. More generally, the certification of FAIR-enabling services is also being studied – a service can enable, respect or reduce the FAIRness of its holdings. 5.3.2. Gaps Existing work on FAIR metrics and certification should be extended under the next framework programme to ensure applicability across disciplines and support implementation. FAIR assessment should be inclusive and progressive, and its usage should take the specific context and needs into account. Several gaps and potential opportunities for extension are noted below: ● Metrics should be combined with a FAIR assessment framework that reflects the needs of different communities while offering comparable methods to assess FAIRness. ● The present checks are good for a proof of concept, but to make general rules for inclusion the scope of the tests was not broad enough; it has to be expanded considerably to explore potential problems and fine-tune the recommendations. ● Different communities attach different weights to the criteria, in particular but not only to interoperability, which has to be fully taken into account. ● The individual assessment models and metrics should be aligned with RDA core metrics and should not hinder a comparative evaluation. - 103 - ● The model can already be used to measure progress on the path to FAIRness, but care should be taken before applying the model for pass-or-fail measurements. ● The need to develop automated evaluation tools for scalability is recognised but there are risks associated with the tool biases. ● Alignment of repository certification schemas with FAIR is under way but needs to be further developed and tested. ● Other critical elements include PID services, semantics and registries, for which assessment frameworks have yet to be defined. ● All the assessment frameworks have to be maintained over time, taking into account feedback from implementation and evolving requirements; the FAIR principles themselves may have to be maintained. 5.3.3. Priorities Significant progress has been made on defining FAIR metrics for data and certification schemas for repositories. This should continue to be built on rather than reinventing the wheel, particularly given the global input and consensus fostered via the Research Data Alliance on these topics. Priorities for FAIR metrics lie in implementation and robustly testing across research communities. For certification of services, support is needed in aligning frameworks with FAIR, developing models for certifying core services such as PIDs, and enabling uptake. Priorities for FAIR metrics Priority 1: Support the assessment and improvement of the RDA FAIR Data Maturity Model. Priority 1.1: Support disciplinary communities to clarify their requirements with respect to FAIR and identify cross-community use cases. Priority 1.2: Test the FAIR Data Maturity Model in a wide range of communities, in a neutral forum and seeking international agreement, to fine-tune and customise the recommendations and guidance, assess the degree of priorities, identify adverse consequences and apply corrections. Priority 2: Assess and test the proposed EOSC FAIR data metrics in a neutral forum, which could be a Working Group set up by the RDA Global Open Research Commons Interest Group, to seek global agreement with the international EOSC counterparts, in addition to any EOSC-specific Task Force or Working Group addressing FAIR metrics. Priority 3: Support the definition and implementation of evaluation tools; their thorough assessment and evaluation, including inclusiveness; comparison of tools (manual, automated); identification of their biases and applicability in many different contexts, including thematic ones. Priority 4: Support the definition of FAIR for software and of the assessment framework for key elements of the FAIR ecosystem, in the first instance PID services and semantics. Priority 5: Define and implement governance of the principles, assessment frameworks and metrics, adapted to each specific case. Priority 6: Provide guidance for and support to implementation: support data and service providers to progress in the FAIRness of their holdings. - 104 - Priorities for FAIR Certification FAIRsFAIR is working on Priority 1, is also active in Priorities 2 and 3 with a set of repositories, and is working on a framework for FAIRness of services. Priority 1: Support the current efforts to align certification standards and assessment schemas with FAIR. Priority 2: Test the proposed schemas in a variety of communities to gather feedback and update the proposed framework accordingly. Priority 3: Provide support, methodologically as well as financially, to data and service providers to progress towards certification. Priority 4: Monitor the progress of certification, assess the maturity of the certification landscape, and take appropriate action if fields or regions are lagging behind. Priority 5: Support the establishment of core criteria and methodology to certify other key elements of the FAIR ecosystem, in particular in the first instance PID services and vocabulary repositories / metadata registries, and test them extensively. Priority 6: Support the establishment and maintenance of registries of certified components of the ecosystem; if several registries are available for a given component, they should be harvestable and included in registries of registries. Priority 7: Establish a Working Group under the EOSC Stakeholder Forum to ensure the implementation and further development of recommendations in the ‘Recommendations on certifying services required to enable FAIR within EOSC’ report [WG_FAIR_CSRecs]. 5.4. Authentication and authorisation infrastructure The purpose of authentication and authorisation infrastructure (AAI) in EOSC is to support the FAIR principles for data and services while enabling high-trust collaborations to be established and maintained with little or no friction to the end user. As federated AAI provides trusted identity information and allows scalable management of roles and rights, it is a key concern for the security and trust of any collaboration. AAI for escience is developed not in a vacuum but in the context of a global marketplace of AAI products and services which typically focuses on the consumer-business relationship. The goal of the EOSC AAI is to build a foundation for e-science AAI which will ensure longterm availability of the aspects of digital identity that are unique to scientific collaborations and which are often hard or even impossible to achieve using the tools and design patterns used to provide enterprise or consumer identity. 5.4.1. Status Fortunately, the e-science AAI community has a long history of building globally viable solutions for digital identity, which can continue to grow and develop within the EOSC framework. The AAI for EOSC can build on a large body of existing work that has been carried out in the Federated Identity Management for Research [FIM4R] activity and the AARC and AARC2 projects [AARC] and its governance spin-off AEGIS [AEGIS], in which a large number of e-infrastructures and research infrastructures are represented. Most notable is the AARC Blueprint Architecture [AARC_BPA], which has been embraced by most large research collaborations worldwide and which describes the components of an interoperable AAI for - 105 - research collaborations. The AARC BPA describes how community AAIs and infrastructure proxies can leverage eduGAIN [eduGAIN], the federation of national research and education (R&E) identity federations and other sources of identity for global science collaboration. 5.4.2. Gaps Despite more than a decade of development in the field of global AAI for the research and education community, a period that has included establishing large-scale global systems such as eduGAIN and eduroam, both the user experience and the service provider experience remain confusing for large parts of the R&E AAI ecosystem. The EOSC effort provides a unique opportunity to address these challenges. To guide this work, the SRIA authors have turned to the first principles of the EOSC AAI: ● User experience is the only touchstone; ● All trust flows from communities; ● There is no centre in a distributed system. From these first principles the following problem statements have been derived: ● There is no consistent user experience for AAI across the e-science ecosystem; ● There is no consistent interface for service providers in the e-science ecosystem; ● The AAI ecosystem must grow to match the growth of EOSC beyond the research and education community. Each of these problem statements is expanded below. There is no consistent user experience for AAI across the e-science ecosystem Currently the user experience for authentication and identification is fragmented. A user authenticating to several services cannot count on any aspect of that behaviour to be consistent, except possibly for the login screen of the home organisation identity provider (IdP) (if the user ever gets that far, that is). In order to successfully identify to a service, a user must: ● Be able to identify the correct gesture to initiate a login flow – in other words, be able to find the login button on the page in the case of a web application; ● Be able to find her login provider (home organisation) among the offered alternatives; ● Have access to a login provider that offers a combination of authentication and identity assurance that matches the requirements of the service; ● Be able to understand what the login process entails in terms of authentication options, credentials, tokens, gestures, etc.; ● Have the appropriate association with the chosen identity provider (employee, student, etc.). In summary: 1. Services must be universally reachable, in the sense that users should be able to either successfully authenticate to all services or understand why they are not permitted access. 2. All participating identity providers must participate in a common framework for managing attributes across the ecosystem. - 112 - stimulates the definition and evolution of adopted interoperability frameworks beyond the community and infrastructure domains. Even if technical challenges are overcome, it is not obvious that, due to legal, organisational and/or financial constraints, researchers are allowed to access a service, data source and/or research product to which they have no direct access. These constraints are commonly set by non-technical boundaries defined in partnership agreements to which a resource provider belongs, in the mandate given to the resource provider or in national legislation and/or regulations. For EOSC to achieve its vision (see Section 2 EOSC in the making) it is necessary to overcome not only the technical but also the non-technical challenges for resource providers to provide access to resources to any researcher within Europe in a sustainable way. The EOSC Rules of Participation Working Group has been given the task of specifying the initial conditions for resource providers to participate in EOSC. The rules are expected ‘to set out in a transparent and inclusive manner the rights, obligations and accountability of the different stakeholders taking part in EOSC’. The EOSC Rules of Participation and Interoperability Framework should contain legal and organisational aspects that allow resource provisioning in a sustainable way across resource providers and across national, community and partnership boundaries. They must also be sufficiently concrete for resource providers to understand how to comply and for EOSC to validate. 5.6.3. Priorities Priorities for this action area are: ● Ensure more efficient onboarding of resources and integration with existing research community catalogues and repositories. ● Enable the composability of resources and across resource providers. This should be optimised and automated as far as possible, for example in resource delivery (e.g. ondemand and self-serve) and composability via the EOSC Interoperability Framework. ● Ensure the guidelines for resource providers are appropriate to and respectful of the existing interoperability frameworks available at a community level. ● Incentivise resource providers to produce and operate resources that are Open Science by design. 5.7. EOSC Interoperability Framework 5.7.1. Status Achieving a good level of interoperability within EOSC is essential to federate data and services and provide added value for EOSC users, across disciplines, countries and sectors. In the context of the FAIR principles, interoperability is discussed in relation to the fact that ‘research data usually need to be integrated with other data’. Standards are critical to achieve this, at both the disciplinary and cross-domain level, and implementation must build on existing research culture and practices, as well as existing technologies such as the Semantic Web, linked data and knowledge graphs. Efforts should also focus on addressing gaps where standards do not yet exist, to avoid the risk of leaving certain research communities behind. Full interoperability, between data sources and services using different standards and semantic artefacts, is difficult to achieve at this point in time, but through EOSC the use of - 113 - standards is being encouraged/required to enable crosswalks and as much interoperability as possible. In addition, the data need to interoperate with applications or workflows for analysis, storage and processing. The EOSC view on interoperability should consider not only data but also the many other research artefacts that may be used in the context of research activity, such as software code, scientific workflows, laboratory protocols, open hardware designs, as well as the services that allow the handling of such data. The current EOSC Interoperability Framework focuses mostly on the digital object level and recommendations are made for expanding this in the next phase of work to address services and other components too. For example, rules for service operation should require a level of reliability and availability to guarantee stable service levels. In terms of EOSC, the ‘I’ of FAIR is the critical aspect, as interoperability is the glue that allows EOSC to function. In order to enable data to be discovered and accessible, a minimum set of metadata, common standards and, preferably, machine-readable semantic artefacts that can interoperate needs to be agreed. Interoperability across countries, data repositories and disciplines is fundamental to the EOSC vision and a prerequisite for the federated approach. Some work has been started in this regard. The FAIRsFAIR project has been reviewing generic metadata standards to recommend approaches for common discovery in EOSC, a co-creation project has been funded to review the DDI-CDI standard and RDA Working Groups are considering standards such as schema.org. A comprehensive review of all possible generic standards to adopt within EOSC must be conducted and in-depth consultation with the full range of research communities must take place to determine which will be most appropriate to apply for broad uptake. 5.7.2. Gaps This section is organised according to the different layers of interoperability that are identified by the European Interoperability Framework: technical, semantic, organisational and legal. In addition, there are three overarching activity areas under which these various gaps and associated priorities fall: ● Support for standards development and adoption EOSC cannot enable FAIR and support interoperability without standards to describe and understand digital objects. Many of the gaps identified address a lack of standards or low levels of adoption, both of which need to be addressed incrementally to enable the full benefits of FAIR to be realised. Once research community standards are in place, work can be performed to map between these, enabling data and services to be used in wider contexts. Turning FAIR principles into practice requires an enormous amount of human skills and support, as well as the standards and technological resources. This gap is even larger if the data coming from the long tail of science are taken into consideration, so work to professionalise data stewardship roles and ensure appropriate levels of support and services are in place is key. ● Engagement with research communities FAIR should be implemented according to the subsidiarity principle, preferencing standards of research disciplines over more generic, less rich metadata. Engagement with professional scientific unions or scholarly societies, research infrastructures, data stewards and software engineers that work closely with research communities and represent their needs at an institutional, European and global level is key to ensure standards have wide applicability and adoption. Fora such as W3C, the RDA and other bodies defining standards at the global level also play an important role here. - 114 - ● Robust governance and implementation A clear governance framework is required for implementation to specify how the different levels of interoperability will be handled across organisations and user communities. Policies from funders and institutions should require and/or incentivise the curation and use of agreed standards. Moreover, recommendations can be made to ensure common services, such as PID resolution, function consistently irrespective of the type of identifier used. At the technical level, the main gaps with regard to achieving better interoperability in EOSC include the following: ● When trying to work with infrastructures or services across communities, authentication and authorisation often needs to be performed separately for each community/service. ● Research data may be made available in multiple general-purpose formats (CSV, Excel, database dumps, JSON, XML, shapefiles, etc.) or community-based models (e.g. Darwin Core, VOTable and VOResource, FITS, NetCDF), which are usually hard to align when reusing datasets across communities. ● Coarse-grained or fine-grained research data from other communities may be difficult to find, given the lack of knowledge about how to query their repositories. ● Multiple service providers for different types of PIDs exist. As a result, different sets of policies are enforced to varying degrees, and sometimes the identifiers are not even resolvable. At the semantic level, many of the interoperability gaps have already been identified in Section 5.2 of this document. The primary issues are as follows: ● Need for principled approaches and tools for ontology and metadata schema creation, maintenance, governance and use. Different communities are using different tools and representation models for their semantic artefacts. Some communities have no agreed standards and no strategies for bridging that gap. ● Need for harmonisation across disciplines. It should be possible for a user from one community to add metadata to existing items (data and semantic artefacts) according to their own research discipline practices (e.g. for a social scientist to add DDI-based metadata to a dataset coming from an environmental scientist). Allow a researcher to transform metadata (or data) from one discipline’s format/annotations to another’s. ● Need to harmonise the same type of data (e.g. observational data in environmental sciences, as is being done in the I-ADOPT RDA WG). ● Need for federated access over existing research data repositories (both inside a discipline and across disciplines). Ability to support discovery of data on the basis of a high-level description, and possibly also on more details such as concepts related to observations and variables. ● Lack of tools for deduplication of legacy metadata records and their quality validation. Use Case: Astronomy – interfacing an existing, global, widely used disciplinary data-sharing framework with EOSC Astronomy has been a pioneer of open data sharing and remains at the forefront. The astronomical Virtual Observatory (VO) is the disciplinary interoperability framework which enables astronomers to discover, find, access, interoperate and reuse the data they need for their research. It is defined at the international level by the International Virtual Observatory - 115 - Alliance (IVOA) and is widely used by data providers worldwide. It provides seamless access to resources and is thus almost invisible to the astronomers but underlies some of the mostused tools. The IVOA Registry of Resources lists more than 100 ‘authorities’ that provide at least one VO-enabled resource. Interfacing the IVOA with EOSC would allow the astronomy community to use resources provided by EOSC as seamlessly as possible, for instance computing resources or EOSC-Core services such as AAI. For example, the IVOA Registry of Resources should be integrated with an EOSC catalogue of services, and the semantic vocabularies provided by the IVOA recommendations should be interfaced or integrated in the relevant EOSC services. Another aspect is to assess how VO-enabled resources could be ‘onboarded’ as EOSC services, in a way compatible with the VO service description and deployment architecture. The integration of VO data and services into EOSC is being assessed in the ESCAPE cluster, which focuses on connecting ESFRI projects to EOSC. The IVOA Registry of Resources, which is built on the Open Archives Initiative Protocol for Metadata Harvesting (OAI-PMH) and includes the Dublin Core plus disciplinary extensions in its metadata schema, has been included in EUDAT B2FIND [EUDAT_B2FIND]. B2FIND exposes the resources and allows filtering by keywords, which works well for the high-level resource description. Figure 5.1 below gives a composite of screen copies of searches for IVOA resources in B2FIND. Figure 5.1: Composite image of searches for IVOA resources in B2FIND This integration exposes the global astronomical data resources in the B2FIND interdisciplinary environment. It can be seen as a first step on the way to federating into EOSC but lots remains to be done as the EOSC structures mature. In order to be accepted and bought into by the astronomical community, any interface should preserve the existing, widely used capabilities provided by the IVOA framework, respect the disciplinary culture, which includes open access to data, and should not require a high overhead from the data providers. Ongoing consultation with research communities is therefore critical as EOSC Rules of Participation, minimum metadata frameworks and interoperability requirements are specified. - 116 - At the organisational level, the following gaps have been identified: ● Need for a clear governance framework that includes clear instructions on how the other levels of interoperability will be handled across organisations and user communities (data formats, AAI services, metadata schemas, ontologies, etc.). ● Need for documents explaining terms and conditions and acceptable use policies for services providing interoperability. For instance, providing clear descriptions of the service-level agreements (SLAs) of those providing catalogues and registries of semantic artefacts, or providing systems to overcome semantic differences between different data sources, or alignments between models. ● Need for interoperability certification mechanisms for service providers, so that service users can set their own expectations about the support for interoperability of those services. At the legal level, the following gaps have been identified: ● Lack of clear statements of rights or information on the legal conditions under which data can be accessed. In effect, much content is shared without a usage licence, let alone a standardised, machine-readable licence. ● Wide adoption of standard open licences for data, code and other outputs to ensure data can be combined without conflicts in licence terms. ● National copyright varies across countries so there is a need for clear licences or a good understanding of how and which type of data can be exchanged, taking into account different jurisdictions. ● Consistent, machine-readable consent agreements to ensure permissions to access or use data are clear and regulations such as the General Data Protection Regulation are met. ● Need for a repository of machine-readable licences that can be associated to different types of research outputs. ● Need for machine-readable schemas for the representation of the main characteristics of service-level agreements. 5.7.3. Priorities As a result of the previous analysis of gaps and needs, the following recommendations can be made to include as priorities for further development of interoperability in EOSC at the technical level: ● Use open specifications, where available, to ensure technical interoperability when establishing EOSC services. ● Define a common security and privacy framework and establish processes for EOSC services, to ensure secure and trustworthy data exchange between all involved parties. ● Define an AAI process for EOSC that is common across communities, easy to implement by resource providers and easy to understand by users. ● Ensure service-level agreements for all EOSC resource providers are easy to understand by users from different communities. ● Enable discovery of data sources available in different formats, either generic or community-based, to facilitate overcoming their heterogeneity, and provide easy access and tools to integrate data across communities, enabling the usage of these data. ● Provide tools for quality validation of metadata records and content of digital objects. - 117 - ● Make available search tools for coarse-grained and fine-grained datasets (and other research objects). There will be a range of general-purpose and domainspecific/specialised search tools, exploiting general-purpose and domain-specific metadata. ● Implement the EOSC PID policy, accommodating any appropriate PID usage, recognising that established practices are at different levels of maturity for different resources and that new PID types may emerge. At the semantic level, the following priorities have been identified: ● EOSC should provide support for the maintenance of repositories of semantic artefacts, and governance frameworks for such repositories, taking into account common practices and stages of semantic resource development and usage of different communities. ● EOSC should define clear protocols and building blocks for the federation/harvesting of these repositories of semantic artefacts. ● Research communities should be well supported (independently of their current state of semantic artefact adoption) so as to generate clear and precise definitions for the terms they use, as well as for their metadata and data schemas (and to incorporate those that they are already using) and their documentation. EOSC should provide support to make these definitions publicly available and referenceable by persistent identifiers for machine readability. ● Urgent, additional resources (financial, but also skills and training) should be dedicated specifically to communities with less developed or no community standards, to mitigate the risk of EOSC becoming inaccessible to the majority of researchers within academic institutions. ● EOSC should propose a minimum vocabulary to allow discovery over federated research artefacts (data, software, publications, etc.) across scientific communities, and based on existing metadata models (e.g. DCAT, DDI 4 Core, DataCite core schema, OpenAIRE Guidelines). There should be some alignment among them, and this vocabulary should be extensible, to allow for disciplinary metadata that is typical for some research communities. At the organisational level, the following priorities have been identified: ● The current set of Rules of Participation recommendations should be completed with aspects related to interoperability. For instance, for data providers this may include asking explicitly that data is published according to specific data formats and/or vocabularies for a specific community. ● The same is applicable to services, which may be recommended to ingest or output data according to such standardised data formats and/or vocabularies, and to their corresponding metadata, with some level of quality. Finally, at the legal level of interoperability, the following priorities have been identified: ● A list of EOSC-recommended licences and their compatibility with Member States’ recommended licences should be provided to data producers, right-holders and users, so as to avoid an inadvertent breach of copyright and with a view to harmonising and reducing the overall number of recommended licences. - 118 - ● EOSC should seek to develop and implement minimum standardised, humanand machine-readable expressions of right statements and use conditions, to be included in metadata and be used by all repositories regardless of discipline. ● Need for metadata schemas for service-level agreements. ● EOSC should consider developing a centralised source of knowledge and support on copyright and licences to users and data generators and to address common Q&A. As noted in Section 5.3 FAIR metrics and certification, these recommendations necessitate investment in the development of data standards, crosswalks and registries to support a FAIR ecosystem. Two key areas of activity for the next framework programme are the support of community standards and the proposal of a minimum vocabulary to allow discovery over federated research artefacts (data, software, publications, etc.) across scientific communities. This section has described the seven action areas relating to the primarily technical challenges and prerequisites to implementing the EOSC ecosystem. The following section addresses the challenges and prerequisites relating to boundary conditions. - 119 - 6 Boundary conditions Seven action areas have been identified by the EOSC governing bodies to help deploy the EOSC ecosystem which relate to the social, financial, legal, educational and cultural challenges and prerequisites to its implementation. These are classified as boundary conditions and are: • Rules of Participation; • Landscape monitoring; • Business models; • Skills and training; • Rewards and recognition; • Communication; • Widening to public and private sectors and going global. This section describes each of those areas, for each one providing an assessment of status, identifying gaps, proposing priorities and, where appropriate, describing some further considerations which should be taken into account. 6.1. Rules of Participation 6.1.1. Status In the current European research landscape, Open Science practices are not yet the norm amongst many researchers. Data and other digital research objects are not consistently findable, accessible, interoperable and reusable (FAIR), and the current landscape of regional, national, European (and international) research data infrastructures (RDIs) is distributed, diverse and fragmented. This presents barriers to the open sharing of scientific results. Research collaboration in Europe and globally can be further increased to realise more and better science if changes are implemented to adopt Open Science practices, make digital research objects FAIR and federate RDIs. EOSC aims to achieve this by stimulating widespread changes in the research environment. The Rules of Participation (RoP) [EOSC_RoP] provide transparent and consistent terms for participation in EOSC, helping to build the trust and confidence required to support this process of change. These RoP are set at a level to encourage wide participation, including from less advanced research communities. The EOSC legal entity, the EOSC Association, will be responsible for the RoP, including their monitoring, enforcement and periodic review and updating, to ensure their impact is understood and that they respond to the requirements of the maturing EOSC. 6.1.2. Gaps The current EOSC has evolved through research and development activities undertaken in a number of projects that have progressed largely independently. Each project has proposed and followed its own work plan. However, these work plans have not been as well coordinated as they might have been. For example, schedules of delivery of services could be more coordinated, onboarding requirements could be better aligned, and greater coordination could help avoid gaps and synchronise on overlaps. Whilst project-level governance can monitor compliance of projects against their own objectives and planned activities, unless those activities are aligned a priori, project-level - 120 - monitoring has little leverage on coordination across projects. It has been observed that it is difficult to build a coherent infrastructure through a collection of independent projects. To build a coherent infrastructure that removes silos and provides integration requires tighter coordination across projects. It is not possible to build a highly interconnected road network with contributions from several partners without joint planning of where roads end in one area and begin in another. Thus, whilst local requirements – whether regional or disciplinary – are best served through local planning, there needs to be wider agreement about where and how the local arrangements will integrate. In addition, there needs to be international agreement about the interfaces across national boundaries. For EOSC, this is partly about architectural standards that enable integration at a technical level, about data standards that enable sharing of data, and also about stakeholder engagement to bring about standardisation of policies, processes and procedures. The RoP will define the policies, processes and procedures required to provide assurance of sustainability, transparency, quality and trust in the practices and services offered through voluntary participation in EOSC. Many of the qualities the RoP may ideally require of EOSC participants are not yet widespread and are not universally available. For example, many digital research outputs are not fully FAIR; many repositories are not yet certified; metadata is not fully standardised and a common metadata framework to support discovery in EOSC has not been defined; many services are not yet interoperable; authentication and authorisation infrastructure (AAI) is not yet globally recognised/interoperable; persistent identifiers (PIDs) are not yet universally assigned and unique. Diversity is a major challenge for defining RoP. The evolution of the RoP will be an iterative process, achieved through dialogue with the community and enforced with the consent of the community. It is also important to develop rules that encourage EOSC users and suppliers in the desired directions for EOSC to achieve its objectives, whilst not imposing requirements that are so onerous as to discourage use of and supply to EOSC. The Rules will also need to reflect changes in the wider environment, such as the development of the GAIA-X initiative [GAIA-X]. In the beginning, therefore, many of the Rules will need to provide encouragement rather than impose strict requirements, but can develop over time to include more stringent conditions. This approach is evident in the proposed FAIR metrics and Interoperability Framework, which define different target levels for each time period, incrementally increasing expected levels of FAIRness and standardisation. 6.1.3. Priorities • The RoP define standards for policy, processes and procedures that provide assurance of quality and trust in the services offered through EOSC. • The RoP apply to all digital resources made accessible via EOSC, including data and services. They define a minimum set of rights, obligations and accountability governing the activities of all those participating in EOSC, such as data and service users, data and service providers, and the operators of EOSC itself. • The RoP assume that the governance structure for EOSC will include a governance framework involving the appropriate stakeholders, which includes a legal entity that will assume ownership of the RoP and provide a decision and revision process for them. - 121 - • The RoP may evolve in the future to incorporate elements arising from the FAIR, Architecture and Sustainability Working Groups (WGs), which are developing recommendations in their respective domains. • The RoP provide a conceptual framework for policies and documents relating to issues such as Terms and Conditions and Acceptable Use Policies. These will need to be further elaborated and reviewed with respect to legal regulations before the RoP are finalised. • RoP are about governance, oversight and authority. Without RoP, EOSC becomes no more than a search engine over an unmanaged collection of resources. • It is essential that there is a framework where RoP can be defined, maintained and enforced. • If EOSC is to be delivered through a programme of projects, far greater control over these projects is required. 6.1.4. Considerations Further considerations that should be taken into account with regard to Rules of Participation relate to: • Registration and discoverability; • Transparent subsidiarity; • Federated services; • Federating services; • Global agreement; • EOSC compliance for external services. Each of these is discussed below. 6.1.4.1. Registration and discoverability EOSC will be primarily a federation of existing data and services where data remain in their current repositories and EOSC provides a means to make those data more broadly discoverable and interoperable. To enable this federation, EOSC must recognise resources, or collections of resources, through registration of those resources in an EOSC catalogue. Participation in EOSC is therefore defined by registration of resources as EOSC resources or in an EOSC-recognised collection of resources. Although somewhat tautological, this definition acknowledges the fact that participation works on a voluntary basis; if and when a provider chooses to register a resource with EOSC, it becomes discoverable and accessible through EOSC. A digital resource is therefore considered to be an EOSC resource if, and only if, it is registered in an EOSC-recognised catalogue of resources. Registration of resources also indicates compliance with the EOSC RoP and use of EOSC branding is available only to registered resources. 6.1.4.2. Transparent subsidiarity While participating as a data provider in EOSC implies commitment to the principles of openness described above, custodianship of the data remains with the data provider. Thus, individual data providers determine the precise conditions under which the data they expose through EOSC may be accessed and used, provided that these do not contradict the underlying - 128 - before transactions between users and suppliers of research data, resources and services can be concluded. EOSC will add value by providing frictionless, easy access to data and related services so that research communities can better connect with suppliers, users and funders. EOSC can also promote a cross-fertilising multi-disciplinary environment where investments can be efficiently leveraged and benefit from economies of scale. Patronage/membership-based ‘learning’ funding models promote the perceived value based on being part of a community and finding help and support or networking capabilities for their members. For example, in EOSC this could mean offering private dashboards to each research organisation through which they can track their consumption over longer periods, allowing them to negotiate better terms with the resource providers. Similarly, resource providers would benefit from continuous interactions with (potential) users, generating a private flow of data and insights to better tailor their future offers. 6.3.3. Priorities A workable funding model for EOSC leading to sustainable funding must be prioritised in the next framework programme. The funding models are currently under-developed, specifically in terms of enabling cross-border use of data and services, which will jeopardise uptake. If services are to be free at point of use, a national/EU funding model must be in place to ensure the costs incurred are recovered by the providers. It is not clear how any transactional model with service charges across borders will facilitate use and it could create an unsustainable overhead and barrier for users and providers alike. At the initial stage, the funding solution needs to be simple and effective, but still compliant within relevant regulations. In-depth studies and piloting such as the following are urgently needed. The following activities have been identified as priorities in order to define a viable funding model: ● Perform cost assessments. ○ Assess cost estimates associated with the EOSC-Core services. ○ Assess cost estimates associated with the full Minimum Viable EOSC (MVE). ● Ensure sustainable financing for EOSC. ○ Develop financing schemes for EOSC. ○ Develop monitoring schemes for the in-kind contribution of members. ○ Develop synergies between national and EC funding streams as well as a higher level of coherence in the funding from different chapters (RTD, CONNECT) of the framework programme, and across the three pillars of Horizon Europe. 6.4. Skills and training 6.4.1. Status In order to leverage the potential of EOSC for open and data-intensive research, a key challenge for Europe is to ensure the availability of highly and appropriately skilled people. The vision of a strong EOSC ecosystem that exploits digital technologies and has data and software at its core necessitates a comprehensive skills and education strategy. Skills and training are indeed essential for mainstreaming Open Science practices in research and thus essential for enhancing its quality and efficiency, leading to more new breakthroughs, sparking innovation and ultimately generating growth in the economy. It is therefore important - 129 - to overcome existing gaps and barriers in the necessary skills and training quickly, to reduce the risk of Europe losing a leading position in Open Science. A sustainable EOSC skills and training strategy must address different professional and research roles as well as their functioning in an organisational or team setting. The diagram in Figure 6.1 shows the various actors in the EOSC ecosystem and describes the main tasks related to each role. The report ‘Digital skills for FAIR and open science’ by the EOSC Working Group on Skills and Training details the digital skills required for those actors to practise or enable FAIR and Open Science [WG_Skills_DigitalSkills]. Digital skills are defined by the OECD as a range of abilities to use digital devices, communication applications, and networks to access and manage information [OECD_BDWC]. In the context of FAIR and Open Science, these skills include an understanding of data, software, tools and frameworks. Workforce capacity development is relevant to individual and institutional actors. Individual-level skills and competences form the basis, but ‘the group as a whole is more than the sum of its parts’. 26 Research data, for instance, require collaboration across different roles and responsibilities. Understanding the EOSC ecosystem and the skills challenge calls for a clear definition of the appropriate profiles required to cover the complete research lifecycle and EOSC added value. Figure 6.1: Actors in the EOSC ecosystem: roles and interactions 26 Angus Whyte, Jerry de Vries, Rahul Thorat et al., D7.3: Skills and Capability Framework, 2018, p. 13 [EOSCpilot_D7.3]. - 130 - 6.4.2. Gaps Lack of Open Science and data expertise Using or developing tools for handling data is becoming an increasingly important part of research. 27 However, at the moment, there are not enough adequately trained people to meet current demand for open and data-intensive science needs, let alone to meet increasing demand and diversity goals. Legal/IPR and data ethics expertise is a challenge even among the FAIR data / Open Science experts, while the research community is not equipped to explore opportunities presented in an interdisciplinary environment. There is also insufficient support for the technological development of ‘FAIR by design’ needed for digital research object acquisition in all the research infrastructures and laboratories (‘smart technologies’). This is a key activity to enable open data, open source software and FAIR paradigms to become a reality on a large scale, and in the near future. Interdisciplinarity, coordinated and coherent approaches to skills and competences building and of education and training provision is another area of concern. There is a need for a baseline approach for data stewardship. Lack of a clear definition of data professional profiles and career paths for these roles Data scientists, data stewards, data curators and research software engineers are some of the different actors needed for the development of data-driven, data-intensive science. It is necessary to provide recognition for these roles and define career paths that make them a viable choice. Although the reliance on the emerging new scholarly data and software support profiles are cornerstone elements in the implementation of FAIR data mandates, a very diverse and uneven picture is seen across Europe. A coordinated and coherent approach is needed to address this gap. Fragmentation in training resources Quality and FAIRness of training and learning resources remains a challenge. Fragmentation of existing training initiatives also reduces impact and there is a need to establish coordination with EOSC. 6.4.3. Priorities To realise this vision of a strong research ecosystem with data and software at its core, EOSC has an important role to play in ensuring recognition of data professionals, propagating Open Science skills by aligning curricula, coordinating the collection of relevant learning materials and influencing strategic agendas at the national and European level. Four key priority areas have been identified, with a phasing plan in line with the development of EOSC: 1. Developing the next generation of Open Science and data professionals; 2. Coordinating training and aligning curricula for students and researchers; 3. Building a trusted and long-lasting knowledge hub of learning materials and related tools; 4. Influencing national Open Science policy for skills by supporting strategic leaders. 27 OECD report: ‘The Digitalisation of Science, Technology and Innovation’ [OECD_DSTI]. - 131 - Priority 1: Developing the next generation of Open Science and data professionals Developing the next generation of data and software professionals necessitates a set of activities which EOSC needs to actively pursue: ● Enhance professional data career paths with appropriate rewards and recognition. ● Develop data skills profiles through community understanding and consensus mechanisms. ● Recognise data skills. ● Provide a quality assurance framework and certification mechanisms for trainers and trainees. ● Facilitate and simplify lifelong learning mechanisms for up-skilling. ● Establish Data Stewardship Competence Centres (DSCC) through the establishment of cross-disciplinary and cross-national networks of experts, leveraging existing international models. Priority 2: Bridging the education gap: coordinating and aligning curricula for students and researchers Researchers are at the centre of EOSC and one of the most urgent priorities is to equip them with the relevant digital skills to practice FAIR and Open Science. Being self-sufficient to work with data is not the same as having self-service data and analytics. No matter how consumable the data is, researchers need to be curious and capable of understanding, questioning and taking the right action based on the insights delivered. This, in turn, improves their experience of and confidence in using data. Use Case: Emerging approaches towards implementing FAIR data stewardship in Europe In the past years, it has become clear that there is a large need for, and shortage of, individuals with Open Science and data stewardship expertise within research organisations. This is evident in all research domains and also transcends the institutional and even the national level. For this reason, a lot of effort has recently been put into professionalising data stewardship, bringing together national and global approaches, and both discipline-specific and discipline-agnostic efforts. However, the implementation of these data stewardship competences, education and certification in higher education institutions (HEIs) and researchperforming organisations (RPOs) is still in an early stage. Substantial work related to data stewardship competences, training and education has already been done in two European countries, which may serve as the basis for the EOSC Skills agenda for data stewardship. In Denmark, national recommendations have been provided to gather evidence for supporting pre-qualifications for data steward education at universities [Wildgaard_2020] and the first Master in Data Stewardship is expected to start in September 2021. In the Netherlands, a community-endorsed data stewardship competency framework was developed and recommendations for national implementation of a national FAIR Data Stewardship Roadmap have been formulated, which can now be taken further by the national stakeholders [Scholtens_2019; Jetten_2021]. The roles and stakeholders in the data stewardship landscape that emerged from both efforts align very well and are shown in Figure 6.2 below. - 132 - Figure 6.2: Roles of data stewards in the data stewardship landscape in Denmark and the Netherlands [DS_Roles] EOSC is in an excellent position to coordinate and to support Member States in building the data stewardship capacity that is needed. Two major complementary actions are required to close the skills gap with respect to data stewardship, and EOSC can play a crucial role in both. First, it is essential to train future data stewards based on an accredited curriculum in Europe’s universities. Second, current professionals in the field need to be (up-)skilled and proper career paths for data stewardship need to be defined and acknowledged. By ensuring research groups access the help of data stewards, research outputs will be more effectively managed and shared, ensuring higher quality, reproducibility and more consistent long-term sharing and reuse. Activities to be considered within the realm of EOSC include: ● Align curricula and training with demand. ● Support communities to tailor generic materials to be more relevant to specific disciplinary and professional practices. ● Support and align with other EU and national programmes (e.g. Erasmus+) to help organisations (e.g. libraries, NGOs) engage and up-skill all levels of researchers (e.g. supervisors and mid-career researchers, teachers in higher education institutions) as well as the public (citizen scientists). ● Explore and align mechanics in rewarding early career researchers for Open Science practices in evaluation processes and awarding efforts with ECTS 28 or other formal certificates. 28 European Credit Transfer and Accumulation System, usually used for students (sometimes PhD level, but not everywhere) [ECTS]. - 133 - ● Build and operate a network of researchers-champions in open science. ● Promote and support advanced learning environments as part of the broader Open Science agenda. Priority 3: Building a trusted and long-lasting knowledge hub of learning materials and related tools A key goal for EOSC is to build capacities to sustain learning corpora for data skills and tooling, with activities that include: ● Develop a quality assurance and certification framework for learning material. ● Devise a common framework for learning pathways for different Open Science and data-related profiles. ● Support the development of an EOSC Knowledge/Education Hub as a set of interconnected and decentralised learning platforms and living repositories for knowledge sharing. ● Facilitate the adoption of open learning environments. ● Promote and support innovative ways of learning by employing creative methodologies and technology for teaching/training at all levels, from awareness to focused expertise. 29 Priority 4: Influencing national Open Science policy for skills by supporting strategic leaders In order for EOSC to achieve its aims it is necessary to ensure both the availability of highly and appropriately skilled people, and a policy environment that supports digital skills for FAIR and Open Science. Steering the skills agenda at a national and European level requires a community of influencers with the ability to affect change within their environments to ensure that policy decisions deliver success. This priority is unique in its focus on influencing national policies. This will help countries to create, update and coordinate their national Open Science and data skills policies and activities. But most importantly, it will create the desired culture shift in the policy-making community, creating a generation of visionaries, able to make the connections with emerging technologies (e.g. AI, HPC) and to quickly adapt curricula and training. It will draw on existing Open Science leadership, ambassador and champion programmes such as those operated by YERUN, LERU, IARU, LIBER, SPARC Europe, OpenAIRE and RDA, etc. These advocates will provide a bridge between Open Science best practices and the policy as influenced by the senior political and administrative leaders targeted by this activity. Its aims will include: ● Leveraging existing Open Science ambassador programmes and champions to provide senior leaders with relevant insights to influence national policy. ● Providing a peer network and forum for senior policy leaders, to enable them to lead transformation initiatives in their country or their organisation. Alignment with the CoNOSC group is recommended here. ● Assisting national stakeholders to align skills initiatives across Europe, within academia but also with other sectors. 29 Example Google game for data science/Machine Learning [What_If]. - 134 - 6.5. Rewards and recognition 6.5.1. Status Present-day rewards and recognition (R&R) systems are largely shaped by governmentmandated national and institutional policies and regulations, but they are also stimulated by the competitive environment in which researchers and research-performing institutions compete for funding and other resources. In addition, the driving forces coming from ranking the results of institutions may lead to stimulations related to the metrics used by the rankers. Many R&R systems currently used by research-performing and research-funding organisations tend to incentivise and reward too narrow a range of academic activities – e.g. publishing in prestigious journals and attracting external research funding – and rely on a limited and often problematic set of evaluation tools (e.g. simplistic publication metrics such as the journal impact factor and the H-index). This leads to unequal appreciation of the various fields of science and hinders knowledge utilisation and the uptake of open science practices [Cohen_2019; EUA_RATOS] and facilitates further competition (and waste of resources) instead of collaboration. Moreover, these R&R systems focus on past performance, often not addressing future potential. Finally, individual excellence is often emphasised over collective excellence or team science. Although guiding principles for good practices in research evaluation have been developed, the practical implementation of these is often lacking. Examples are the Leiden Manifesto [Leiden_Manifesto] presented in 2015, and a recent initiative from the Declaration on Research Assessment (DORA), summarising five design principles [DORA_1] that help institutions experiment with and develop better research assessment practices. The DORA initiative maintains a curated list of good practice examples of implementation from institutions showing leadership in this area [DORA_2]. Other initiatives re-evaluate the whole process of research assessment, such as the Association of Universities in the Netherlands (VSNU) position paper ‘Room for everyone’s talent’ [VSNU_PP]. 6.5.2. Gaps To advance from the status quo, a culture change needs to be realised as current R&R systems poorly reward and recognise not only Open Science and FAIR practices, but also aspects of education, research, impact, science communication and leadership. More than a technical issue (e.g. ‘better indicators’), a responsible R&R system is also a social issue: a catalyst to foster good research practice and quality in terms of content, openness, scientific integrity and contribution to society. Future evaluation of researchers should have a better balance in valuing achievements in education (where appropriate); research; influence (on science, society, economy, environment and teaching); organisation and leadership. In particular, and of direct relevance to EOSC, recruitment, evaluation and promotion criteria should recognise openness and FAIR practices, as this is vital for advancing the good practices in research data management that underpin EOSC [CESAER_WP_RDM]. In addition to the evaluation of researchers and research there is an ongoing discussion in the community about ‘the third space’ – ‘the in-betweeners’ – people who do not fit the traditional categorisation of support staff vs. academics, in roles such as data stewards, data managers and research software engineers. They often have a research background (many have PhDs and postdoctoral experience) but have different skills and choose different career paths. However, they clearly contribute to the team science’s collective success, albeit in a different - 135 - manner. This divide between support staff and academic staff affects the perception of people and the contributions they make. It may, however, also result in issues such as difficulty with residence permits for support staff, difficulty with visas, tax issues, inability for support staff to apply for grants, etc. This is an issue not only for Open Science. However, the new professions resulting from the development of Open Science increase this dilemma. All these developments are connected to the broader topic of supporting and rewarding modern research (and research support) careers, where further gap analysis combined with practical guidance and best practice examples are available from a range of sources. Notable examples include the advice paper ‘Open Science and its role in universities: A roadmap for cultural change’ [LERU_AP24_OS], the position paper ‘Room for everyone’s talent’ [VSNU_PP], the Dutch ‘Strategy Evaluation Protocol 2021-2027’ [VSNU_SEP], and the white papers ‘Sharing Experiences with the Human Resources Strategy for Researchers’ [CESAER_WP_HRS4R] and ‘Boost the Careers of Early-Stage Researchers’ [CESAER_WP_BCESR]. 6.5.3. Priorities Overall, two sets of priorities can be distinguished: internal priorities for organisations engaging directly with rewards and recognition systems (including research-performing and research-funding organisations) and external priorities for organisations that set broader rules and framework conditions (e.g. EU institutions and national and regional governments). For internal priorities, several broad lines of action can be identified, including: (i) demonstrating leadership in enacting change towards a culture of trust, openness and risk taking; (ii) preparing Human Resources (HR) to adjust R&R structures (including approaches to recruitment and promotion) using (iii) next generation and progressive metrics [CESAER_WP_NGM]. For external priorities, broad lines of action include: (i) safeguarding institutional autonomy; (ii) identification of legal obstacles; and (iii) removal of legal obstacles to empower researchers, research-performing organisations and research-funding organisations to experiment with, develop and refine better R&R systems. Examples include ensuring sustainable funding levels and that labour laws, migration rules, social security schemes and pension systems support, and do not hinder, modern and diverse careers within academia. Organisational rankers (e.g. those who publish ‘best university’ lists) is a category that falls between the two above, as they themselves rarely interact directly with researchers and their R&R systems, but they also do not have the formal mandate of governments. Nevertheless, rankings have an outsized impact on the behaviour of the research community, and EOSC should engage closely with rankers to ensure that any influential rankings that still use narrow and outdated criteria embrace instead a modern and diverse understanding of what makes excellent research and research teams. Moreover, their handling of data is in most cases not in line with sound scientific practices (cf. the findings of the INORMS Research Evaluation Working Group [INORMS_RtR]). EOSC, as one of the organisations stimulating Open Science, can help in modernising R&R systems aligned with the priorities outlined above by direct action and by promoting action from – and working with – other actors. EOSC should work with researchers, research-performing organisations and their leadership to: - 136 - ● Change the narrative around careers in academia that currently focus on individual competitiveness to focus instead on collaborative and team-based approaches, including the importance of promoting equality, diversity and inclusion; ● Embrace the five design principles outlined in the one-page DORA briefing paper [DORA_1] to help institutions experiment with and develop better research assessment practices: (i) instil standards and structure into research assessment processes; (ii) foster a sense of personal accountability in faculty and staff; (iii) prioritise equity and transparency of research assessment processes; (iv) take a big picture or portfolio view towards researcher contributions; and (v) refine research assessment processes through iterative feedback; ● Embrace a culture of quality, trust and risk taking and reduce the focus on ‘narrow metrics’; ● Involve early-stage researchers in scientific leadership and governance, for example as outlined in the white paper ‘Boost the Careers of Early-Stage Researchers’ [CESAER_WP_BCESR]; ● Follow best practices such as the examples in the white paper ‘Sharing Experiences with the Human Resources Strategy for Researchers’ [CESAER_WP_HRS4R]; ● Sign and /or implement the Declaration on Research Assessment (DORA) [DORA]; the Leiden Manifesto [Leiden_Manifesto] and the Hong Kong Principles [WCRI_HKP]. EOSC should work with organisational rankers to: ● Make their methodologies and data-wrangling processes more transparent, verifiable and thus reproducible; ● Reduce the focus on ‘narrow metrics’ to promote a culture of quality, trust and risk taking; ● Change the narrative in rankings around ‘excellence’, which is currently often defined based on individual competitiveness, to focus instead on collaborative and teambased approaches, including the importance of promoting equality, diversity and inclusion. EOSC should work with the EU institutions to: ● Make an inventory of legal obstacles to improving researcher careers and mobility at the European, national and regional levels; ● Empower individuals, universities and other research-performing organisations to challenge and remove barriers; for instance, following the example of the copyright retention strategy of cOAlition S [Plan_S]; ● Provide support for pilot projects to develop new approaches to evaluating academic performance and to improve staff employability (especially for early-stage researchers); ● Promote wider transformation of national frameworks, using EU funding programmes and various soft instruments as a ‘lever’ to promote open, transparent and merit-based recruitment processes and recognition for all aspects of academic work; ● Create interaction between countries on the topic of R&R systems to facilitate learning and sharing of good practice. EOSC should work with regional and national governments to: ● Produce at least a country-level inclusive approach to research evaluation; - 137 - ● Provide resources and sustainable funding levels to ensure attractive, stable and stimulating working conditions for researchers, teachers and support staff; ● Direct national funding agencies to adopt best practices in modern research evaluation (examples include the Leiden Manifesto [Leiden_Manifesto], DORA [DORA] and the Hong Kong Principles [WCRI_HKP]). 6.6. Communication The Executive Board established a Task Force on Communication to provide clarity on the why, how and what of EOSC, and to set up these messages in a consistent way. The Communication Task Force consisted of members coming from the Executive Board, Governance Board, and communication experts from the EOSC Secretariat, European Commission and elsewhere. There are three areas of importance in communications: stakeholder engagement, content production, and branding and positioning of EOSC for the different stakeholders. 6.6.1. Status There is a toolkit for communication [EOSC_Comms_Toolkit], including a new template for (PowerPoint) presentations, using the current logo. The logo can be used by EC, EOSC and projects and initiatives that work on EOSC. There is a protocol for the use of the EOSC logo by (EC) projects and external entities [EOSC_Brand_Guidelines]. This policy was approved by the Executive Board, and published on the website of the EOSC Secretariat. A standard presentation on explaining EOSC for different stakeholders is also available. Nine different groups of stakeholders have been distinguished [EOSC_Landscape]. For communication purposes, these groups can be aggregated into three main stakeholder groups: ● Research Service Providers: ○ e-infrastructures, such as PRACE, GÉANT, OpenAIRE, EUDAT, EGI, also referred to as delivering horizontal services; ○ Research infrastructures, such as ESFRIs, also referred to as delivering vertical or thematic services; ○ Data and research initiatives, such as RDA, offering global platforms for sharing expertise; ○ Cloud providers, including commercial parties such as Amazon, offering services to research; ○ Cloud community. ● Research Performers: ○ Research communities; ○ Research-performing organisations. ● Research Funders: ○ Research funders; ○ Policy makers. In a later phase, these will be expanded to include citizens, small and medium-sized enterprises (SMEs) and other societal groups as users and enablers. 6.6.2. Gaps The following gaps and requirements have been identified: - 144 - ○ Stimulate and reinforce national top-down initiatives for the promotion of research, with bottom-up approaches by diverse citizen scientist and developer communities. ○ Promote Open Science success stories as a way to support the widening of EOSC. ○ Secure support of Open Science by national governments and funding organisations. 6.7.2. Going global 6.7.2.1. Status As noted in Section 2.6 International dimension, EOSC operates in a global ecosystem with the clear aim to promote the ‘Open Science, Open Innovation and Open to the World’ principles in its international activities. Around the world, regional and national Open Research Data Commons and Open Science Clouds are being developed and several important international policy agreements and initiatives have been established that demonstrate the importance of international cooperation in Open Science. These include the G7 Expert Group on Open Science [G7_OS], the OECD Principles and Guidelines for Access to Research Data from Public Funding [OECD_ARDPF], which is currently being updated, and the Research Data Alliance [RDA] and GO FAIR [GO FAIR] initiatives. The common vision embodied across these international developments enables Europe to enhance scientific cooperation and collaboration with other parts of the world and drive a cultural change towards Open Science based on agreed principles. 6.7.2.2. Gaps Through global cooperation, Open Science has the potential to effectively address many new scientific questions, as well as revisting some long-standing problems. This is particularly true for a number of pressing contemporary challenges. The following are areas where international cooperation is of particular importance: ● Activities with special relevance to complex societal challenges such as climate and sustainable development goals; ● Issues of scarcity such as limited and sporadic amounts of data (e.g. rare diseases); limited availability of the research subject (e.g. rare-earth elements or metals); or a small talent pool in a unique research field (e.g. ITER, black holes, etc.); ● Research fields where the talent pool is very dispersed (e.g. Arctic research); ● Screening for unique solutions developed by local communities, such as indigenous groups; ● Scientific observations resulting from synergies between enabling technologies (e.g. from sensor to satellite such as oceanography). 6.7.2.3. Priorities For the above issues to be addressed, and the full potential of EOSC and Open Science to be realised through a global approach, the following priorities have been identified, taking into account the need to adapt to and consider diverse capabilities and demands, and the principles outlined in Section 2.6. ● Promote an international Open Science culture and the need for change in the reward systems to support the transition of other world regions towards Open Science, where certain regions with less developed research ecosystems could leapfrog. EOSC - 145 - members, especially infrastructures with already existing international cooperation, are particularly suited to address this. ● Initiate an international data steward network across domains to exchange best practices and success stories. ● Promote the uptake of the building blocks of EOSC (such as the EU ICT technical specifications [EC_ICT_TechSpec], the rolling plan for ICT standardisation [EC_ICT_Standard], FAIR, PIDs, AAI, APIs, CoreTrustSeal, etc.) and open source solutions abroad, given that formal standardisation is difficult in the current fastchanging, open source environment of research. ● Promote the EOSC service portfolio abroad, such as the EOSC-EarthOb, which will enable the use of Copernicus and Galileo data more easily, particularly relevant for third countries. ● Provide state-of-the-art trainings on technical requirements of the Horizon Europe calls, such as data management plan (DMP), FAIR, Open Access, to enhance thirdcountry participation and success in Horizon Europe calls. ● Initiate EOSC Rules of Participation (RoP) for service providers from third countries, noting that compliance with applicable legislation is a prerequisite beyond the RoP. ● Develop value propositions to third country service providers, to widen the EOSC portfolio. ● Encourage emerging regional Open Data Commons in countries/regions with commitment to Open Science, eligible for the EU Development Funds. ● Initiate partnerships via Memoranda of Understanding (MoUs) with other Open Data Commons that enable users of each initiative to access the resources of the others. Cooperation with these initiatives should be found at an institutional level, to establish a level playing field, and enable a good user experience. ● Propose a Global Open Data Commons Charter which paves the way to a Global Open Data Commons. This should be developed in close cooperation with the RDA Global Open Research Commons Interest Group, which brings together a number of actors from the relevant initiatives. ● Give support to and cooperate with existing international projects and initiatives, build on their work and contribute to their mission (e.g. Data Together which is comprised of CODATA, GO FAIR, RDA & WDS). ● Enable the formation of international consortia for Horizon Europe calls. ● Systematically embed the sustainable development goals (SDGs) into the EOSC Annual Work Plan and activities, as well as the overall strategic goals of the Horizon Europe programme. Sections 5 and 6 have described key action areas involved in deploying the EOSC ecosystem. The next section considers the benefits that are expected to result from its deployment. - 146 - 7 Expected impacts The climate crisis, the extinction of species, global poverty and social inequality are only a few of the challenges that humankind is facing in the 21st century [EC_HE_Missions]. Research plays a crucial role in addressing these challenges and, against this background, EOSC will be a major European vehicle for joining forces to help transform individual research efforts into collective efforts. EOSC will also help to fill infrastructure as well as social gaps in unstructured areas, and play a significant role in raising to the most advanced level the science domains that have unsatisfied e-needs, with the target to increase levels of integration. Recalling the Objectives Tree presented in Section 3, the final row of the tree describes the benefits of EOSC for the three dimensions of Science, Industry and Society: Figure 7.1: EOSC Objectives Tree – benefits This section considers the impact of EOSC on each of these areas. Section 7.1 addresses the impact of EOSC in improving trust, quality and productivity in science; Section 7.2 looks at the development of innovative services and products; and Section 7.3 discusses the role of research in addressing societal challenges. To conclude, Section 7.1 summarises the critical success factors that must be in place for these benefits to be realised. 7.1. Improved trust, quality and productivity in science Encouraging collaboration and openness EOSC will stimulate the cultural changes in the entire research ecosystem. Open Science, which is realised with the help of EOSC, is striving for better horizontal and vertical links - 147 - between scientists, scientific institutions, research and data infrastructures, and interconnecting scientific disciplines. It equilibrates the traditional research outputs, such as publications, patents, etc., with other forms of research outputs, including, for example, data, software, including models, simulations and methodologies. Making these outputs as findable, accessible, interoperable and reusable (FAIR) as possible is therefore a key requirement in measuring and rewarding the contribution of research. Open Science and EOSC will have a significant structural effect with the potential not only to change the way research is performed, by creating a pan-European, multi-disciplinary federation of research infrastructures supporting a broad range of a researcher’s data and computing needs, but also to enable new mechanisms for communication and evaluation of research, motivating researchers, institutions and national research systems to open their research outputs. Trusted frameworks for data availability and security The foundational fact – indeed, prerequisite – that EOSC provides a secure, safe and transparently trusted virtual environment where scientific outputs can be deposited and found according to the FAIR principles, represents a significant change that will impact the overall quality of research. It unlocks the full value of research and, by developing certified services and standards, will enhance the quality of knowledge management, data discoverability and reuse. EOSC will also underpin the development of new ways to deal with open access to all forms of research outputs, with automated access guided by clear and transparent Rules of Participation that ensure trust in the quality of data and the function of data access services. Researchers will therefore be able to make their data open in the knowledge that their work will be acknowledged, their intellectual property (IP) will be protected where appropriate, and that sensitive data will also be appropriately protected where necessary. Infrastructure planning Alongside the direct impact on science, EOSC will also contribute to the quality of research by reducing the disparities in the Open Science readiness in different countries, reducing the divide across regions and mobilising important resources that will federate national data systems, enabling new actors to foster data interoperability with a high level of interdisciplinary research. The pan-European EOSC will also positively influence the planning of institutional and national infrastructures by developing synergies and compatibility schemes with other existing infrastructures, improving the quality of the integrated research landscape, increasing researchers’ ability to provide science-based solutions to complex societal challenges. Broadening discoverability EOSC will facilitate integration not only within scientific domains but also across domains, offering a trusted and stable ecosystem for linked Open Science. Even within their own field, researchers face challenges in discovering, locating, accessing and reusing relevant data. EOSC will address these challenges in two ways: first, by making data FAIR to enhance discoverability; and second, by federating research infrastructures so that relevant datasets and thematic services from particular fields are more widely exposed, encouraging multidisciplinary research. Making new connections EOSC will enable the ‘intelligence’ and processing power of machines to be utilised to uncover connections and related relevant material that may not be put together otherwise. Metadata - 148 - is a central tenet of FAIR. All digital objects require persistent identifiers and rich contextual information to enable discovery and reuse. EOSC will provide a context where this metadata can be standardised in machine-readable formats so it can be processed at scale by computers, thereby alleviating some limitations of human searching and maximising the potential of machine searching. Addressing global challenges Societal and global challenges demand cross-disciplinary research, and thus datasets from different disciplines must be interoperable. By federating scientific data infrastructures and overcoming fragmentation, access and reuse of data will become easier and more efficient. EOSC will integrate the landscape of research data repositories in Europe, which is currently highly fragmented. By far the largest part of the relevant research data are not stored in repositories. For those data that are stored in local, institutional or disciplinary repositories, they form disconnected research data silos where data are largely unfindable, thus inaccessible and definitely not interoperable. This clearly hampers data reuse, knowledge circulation and, more importantly, it reduces significantly the impact science could have on society in the broadest sense. Example: Addressing the COVID-19 Pandemic When addressing global challenges, multiple streams of data from different fields are needed. COVID-19 is a case in point. To address the pandemic, epidemiological data to track the spread of the disease, understand patterns of transmission and support contact tracing were naturally at the fore. The various applications released to gather these surveillance data raised many social and ethical questions about appropriate access and reuse, requiring strong governance controls and robust authentication and authorisation infrastructure (AAI). Person-level clinical data on patients, such as virology test results and imaging data such as lung scans, as well as sequence and metabolomics data were also needed. To implement effective policy measures, these medical data need to be combined with a much wider range of inputs such as realtime travel information, economic analyses and social insights into likely public responses to proposed measures. The European COVID-19 Data Platform, coordinated by the European Commission (EC) and European Molecular Biology Laboratory (EMBL), enables the rapid collection and comprehensive data sharing of available research data and tools on COVID-19 from different sources for European and global research communities. Practically, this enables researchers to upload, access and analyse COVID-19-related reference data and specialist datasets. A data portal provides the primary entry point into the functions of the data platform, which in turn forms an entry point into the future EOSC. Expanding the European COVID-19 Data Platform to enable the integration of molecular research data with patient and clinical data will ensure that patients benefit directly from the research supported by EOSC. Cross-linking with socio-economic, societal response and other social science and humanities will promote an integrated understanding of European outbreak response and preparedness and demonstrate the value of FAIR data to society and public engagement. - 149 - Through the federation of data and research infrastructures, EOSC will enable the creation of new opportunities and solutions in key thematic sectors such as health, food, transport or environment. EOSC will allow researchers from different countries and disciplines to verify, combine and build upon existing scientific data, addressing questions that cannot be addressed in isolation. In order for EOSC to achieve these goals, there is an onus on researchers to adopt relevant community standards and for the curation community to develop crosswalks for interoperability. Research communities need to be supported to define and adopt data standards, sharing agreements, services tools and know-how to facilitate the reuse of data. Some, such as astronomy, life sciences and linguistics, have self-organised, but many others require support to avoid widening the gap between the research communities active in EOSC or the range of content and resources that are available for multi-disciplinary reuse. Enhancing reproducibility Reproducibility of research results is an essential aspect of research. It encourages objectivity and self-correction as well as discouraging scientific misconduct and fraud. However, it is widely recognised that today many research results are not reproducible. Opening up research processes and outputs is an important way to aid reproducibility. This is true not just for data, though this is critical, but also for all the processes and tools used in the research lifecycle, including software, methodologies, instruments, simulations, and analysis and workflows. EOSC will provide researchers with the means to access complete datasets and analysis platforms and provide services that support reproducibility, as well as ensuring long-term preservation and long-term availability of these research data and tools. Reproducibility also requires a stable and trustworthy IT infrastructure, which is not, in general, provided by an individual researcher’s desktop folders and analysis codes, often developed for one-off use. Where research is undertaken by large teams, this may exist already so that the team can work together. Where research is undertaken by an individual or small team, this is less often the case. EOSC will provide a sustained and stable infrastructure for research, with a multitude of readily available research datasets and tools, thereby encouraging researchers to develop their own research environment on this platform, encompassing reusing existing components, rather than building one-off, non-reusable tools in their own personal IT space. 7.2. Development of innovative services and products Europe is undergoing a digital transformation in all sectors to foster innovation. In science, EOSC will lead to a fundamental revolution in the way researchers, companies and government agencies share and exploit research outputs, somewhat similar to how the internet revolutionised the sharing and exploitation of information. Ultimately, each and every scientist will do research differently from the way it used to be performed. Within the scope of the Co-programmed European Partnership, EOSC will also address the differences in economic development in the research and innovation sector by creating more equitable access to data and services from both users and providers. Researchers and innovators will be able to jointly create innovative new technologies and services, which in turn will lead to the creation of new jobs and markets. Opportunities to improve support for researchers The implementation of the EOSC ecosystem will enable European research to make its digital transition while ensuring transparency, reproducibility and societal impact. By providing - 150 - seamless access to increasing volumes of research data, EOSC will stimulate the uptake of different services, from both public and commercial providers, that align with the principles of EOSC. By enabling access to data and services at European level, EOSC will facilitate and widen the opportunities for researchers to collaborate, and will enable them to start new research activities in their home country without relocating. EOSC will therefore further strengthen a balanced and fair ‘brain circulation’ and achieve a more symmetric mobility of researchers. When the ecosystem of new tools and services is available, and as many new FAIR-by-design datasets are generated as possible, researchers will be able to deliver much more rapidly the outputs of each part of the research lifecycle, including data and software, with the same level of precision as they deliver publications today. For research teams and laboratories, publications, data and software will be managed in a holistic, synergistic way, as interrelated digital objects, in order to optimise the reuse of research results. The EOSC Web of FAIR Data and Services will provide the ideal ground for building a wide range of new innovative and value-added services (from visualisation and analytics to longterm preservation). It will be as transformative as the World Wide Web has been to business and everyday life. The consolidation of (FAIR) data commons and the interconnection of research data silos will also enable the creation of new opportunities and new solutions in key thematic sectors such as health, food, transport or environment. To encourage the development of innovative services supporting FAIR principles, as well as data stewardship and preservation across different phases of the research lifecycle, dedicated incentives schemes funded by the EC are foreseen that would use the EOSCExchange as a distribution channel. For example: ● Research and Innovation action grants to develop services to be made available via the EOSC-Exchange; ● Pre-Commercial Procurement / Public Procurement of Innovation Solutions (PCP/PPI) co-funding financial instrument for innovative services to be co-developed with the private sector, procured jointly by public authorities and commercialised via the EOSCExchange. All such innovation incentives would require developments to adhere to Rules of Participation resulting in production-quality services (Technology Readiness Levels 7–9) to be included in the EOSC-Exchange with associated training material. Opportunities to improve support for the private and public sector EOSC will enable the additional functionalities and services that it provides to serve not only the research community but also the public and the private sector so that they can exploit open data and associated services in such a manner that greatly increases the potential for innovation and economic impact in Europe. EOSC will bring more actors and investments into the research and innovation process. EOSC will be instrumental in stimulating many areas of the European private sector, for example, the cloud and artificial intelligence (AI) industries, that are willing to align to these principles while, at the same time, it will ensure that European researchers remain in control - 151 - of their data, stored in trusted and FAIR-certified European repositories, and that scientific knowledge will stay ‘as open as possible, as closed as necessary’ 31 . Example: The Copernicus Data and Information Access Services One inspirational example is the Copernicus Data and Information Access Services (DIAS), which provide access, tools and processing capabilities for scientists and innovators to exploit Sentinel data. The five DIAS online platforms are operated by the industry and allow users to discover, manipulate, process and download Copernicus data and information. All DIAS platforms provide access to Copernicus Sentinel data, as well as to the information products from Copernicus’ six operational services, together with cloud-based tools (open source and/or on a payper-use basis). Federating Copernicus data and DIAS added-value services into EOSC will leverage the existing EC investments for the benefit of multiple science and innovation communities. This will reduce the burden on scientific institutes to engage in complex procurement processes, support cross-analysis of data from heterogeneous sources, create market opportunities for research data services and represent a demand-side stimulus for the commercial DIAS. Opportunities to improve European leadership and collaboration in a global setting The EOSC Partnership will increase European leadership in Open Science and provide opportunities to strengthen international cooperation. EOSC has begun as a European initiative, federating research data repositories and infrastructures across Europe, but the ultimate goal of EOSC is to lead the development of a Global Open Research Commons, of which EOSC will form the European component. EOSC will be European and open to the world, reaching out over time to relevant global research partners and initiatives so that by 2027 there can be alignment and interoperability of infrastructures to promote Open Science globally. Coordination fora including COAR, CODATA, RDA and WDS [COAR; CODATA; RDA; WDS] provide an environment where the different layers of interoperability (legal, organisational, semantic and technical) 32 can be discussed with partners from around the world. There is a clear willingness to collaborate and it is expected that the first agreements will be put in place during the first iteration of EOSC. 7.3. Improved impact of research in addressing societal challenges Research in society Through the introduction of EOSC, research will gain public awareness and will meet the public need to trust scientific facts. Against this background, empathy, transparency and the mediation of research ethics will have as big an impact on the public status of research as will data quality or quantitative ways of measuring impact, whilst both concepts will enhance societal resilience and meet socio-economic needs. EOSC will make possible a much higher 31 For research artefacts, the interpretation of “as closed as necessary” is “as restricted as necessary”. 32 Layers of the interoperability model defined in the European Interoperability Framework [EC_Interoperability]. - 152 - level of interdisciplinarity and scientific evidence in decision making, planning and strategy at societal level. EOSC will help Open Science to become the new normal. EOSC envisions a sustainable and federated infrastructure that offers standards, tools and services, allowing researchers to find, access, reuse, and combine scientific results, and in which these researchers are trained and rewarded for Open Science. This will improve the quality and productivity of science, with researchers being able to access and exploit other research as well as collaborate with other researchers, and will increase public trust in science as an open and evidence-based enterprise for society. This renewed trust in science is crucial given the rise of fake news and loss of trust in experts. This will also stimulate the development of innovative services and products arising from scientific breakthroughs, further stimulating scientific advancement and fuelling the economy by stimulating market competition, creating jobs and encouraging consumer spending. These objectives and benefits, in turn, improve the impact of research in addressing the global societal challenges of the times and give a return on the public investment in science. Supporting international collaboration As the COVID-19 pandemic has dramatically shown, immediate and open access to scientific research is crucial to deal with urgent societal challenges. EOSC will ensure that scientific publications, data and software relating to urgent societal problems are discoverable, accessible and reusable for other researchers to speed up breakthroughs, such as finding a solution to halting the spread of and ultimately vaccinating against COVID-19. Better and faster sharing of research will naturally strengthen collaboration among researchers and disciplines as well as create opportunities for new levels of integration. The interoperability of data will also lead to unexpected links across disciplines as well as stimulate and support multi-disciplinary research. EOSC will, in effect, bring researchers within and across disciplines together and help science become more of a team enterprise. This is crucial for successfully tackling large-scale societal challenges, such as the Horizon Europe missions, which typically involve complex problems and require solutions from a multitude of different disciplines. One example is climate change, which is a truly multi-disciplinary research domain and can include botanists, climatologists, computational modellers, geochemists, mathematicians, meteorologists and oceanographers. While these researchers need to find one another and learn to work together, they also need the right tools to be able to collaborate effectively. EOSC will provide a catalogue of value-added services that will provide computation, storage and analysis as well as other data-related services and tools to help researchers collaborate in a multi-disciplinary environment. Lifting science beyond the human scale For science to really break boundaries, researchers need to think beyond what they currently know and make connections that they do not currently see. One barrier is the exponentially increasing amount of data being produced, which is already too much for a human to process. Another barrier is the lack of interoperability across datasets, resulting in a fragmented data landscape. A further barrier is that humans are not able to pinpoint statistical correlations across a diverse range of different disciplinary datasets in a reasonable amount of time. EOSC will lift science to a new technological level and help researchers make discoveries that could never be made with conventional methods. The deployment of smart algorithms, machine learning and AI services onto the Web of FAIR Data will allow unexpected - 153 - correlations to be made across all interconnected datasets in real time. It is then the researchers’ task to investigate these new scientific avenues and determine causation from the correlations, and the innovators’ task to convert this new knowledge into societal benefit. Imagine, for instance, running a search on ‘malaria’ in a research discovery portal that has access to the Web of FAIR Data. Within seconds, the search tool delivers a structured collection of results summarising all related articles and relevant data from both expected and unexpected sources (such as a climatological institute), industrial stakeholders (such as a pharmaceutical company), and public institutions (such as a hospital). And then, after a short interaction to understand the nature of the enquiry more precisely, the search tool suggests a specific treatment for specific patients in a specific region: an exciting potential discovery only made possible through an approach such as EOSC. 7.1. Critical success factors The developments and expected impacts described above will not happen spontaneously. For these benefits to materialise a number of critical success factors (CSFs) must be in place. The following CSFs have been identified for EOSC: ● Researchers performing publicly funded research make relevant results available as openly as possible; ● Professional data stewards are available in research-performing organisations in Europe to help implement FAIR principles and support Open Science; ● Researchers are skilled and incentivised to perform Open Science; ● The scope of EOSC is widened to serve the public and private sectors; ● Research data produced by publicly funded research in Europe is FAIR by design; ● The EOSC Interoperability Framework supports a wide range of FAIR digital objects including data, software and other research artefacts; ● European research is increasingly discovered and reused across disciplines as a result of EOSC; ● EOSC is operational and provides a stable and trustworthy infrastructure, supporting researchers addressing societal challenges; ● EOSC has a sustainable funding and business model; ● EOSC is populated with a valuable corpus of interoperable data and services; ● The provision of data and services is available across borders; ● EOSC is a valuable and valued resource to a wide range of users from the research and education, public and private sectors. To manage these prerequisites, some have been translated into activities, outcomes and key performance indicators in the EOSC roadmap; others will be the focus of a sustained stakeholder engagement and communications strategy. The next section outlines the roadmap for delivering a fit-for-purpose EOSC that will in turn deliver the expected impacts described above.