Expressing scholarly information primarily in narrative documents is outdated and hindering scientific progress. The use of printed articles (and their now pseudo-digitalized static PDFs) is a relic of historic developments dating back to the very beginning of science (see Fig. 1). In the age of advanced information systems powering fast-paced knowledge economies that face numerous global societal challenges, it is no longer adequate to express scholarly information - an essential resource for modern economies - primarily as article narratives in document form. Text and data mining tries to overcome these issues, with only limited results, unless the extracted information and data are liberated, quality checked and FAIRified to allow for a wider and efficient reuse.

Figure 1.  

While other information-intense domains were completely disrupted (e.g. mail order catalogues, street maps), scholarly information continues to be published as static documents, a root cause of major challenges for the scientific method overall, including proliferation of publications and reproducibility crisis. (Sources: 17^th century, https://doi.org/10.1098/rstl.1665.0001; 19^th century, https://www.biodiversitylibrary.org/page/39489576; and 21^st century, https://doi.org/10.5334/dsj-2020-018.)

The human effort required to comprehend information expressed in such form can no longer keep pace with the overall speed of science and subsequent demands on all research lifecycle phases, including information production and review. The urgency of results, seen for instance in vaccine research among other global societal challenges, or the relentless growth of scholarly information, makes it increasingly hard to gain or maintain an overview of the state of the art (Jeschke et al. 2019). The current lack of FAIRness of scholarly information is a serious impediment to science productivity, and the more efficient use of scholarly information is an important current need of the research community at large and, more broadly, the global society. We argue that the need to more efficiently use scholarly information will only grow stronger in the future, yet will be hindered by the growing number of documents added to the total.Hence, the European Open Science Cloud (EOSC) and its infrastructures urgently need to increase their service offering and capabilities to address the FAIRness of scholarly information. Indeed, the 2016 first report and recommendations of the Commission High Level Expert Group on EOSC (European Commission 2016) had already identified “New modes of scholarly communication (with emphasis on machine actionability) need to be implemented” as a key factor for the effective development of the EOSC as part of Open Science. In 2018, the European Commission expert group on FAIR data provided a follow-up report (European Commission 2018) including an action plan with recommendations for stakeholders to take towards a FAIR ecosystem, calling for discipline-aware implementations of EOSC-related infrastructure centred around FAIR Digital Objects, including scholarly information. Similarly, the 2020 EOSC Secretariat Workshop (Akgun et al. 2020) on “Co-creating the EOSC: Needs and requirements for future research environments” underscored the need for services for machine actionable scholarly information sharing. Below, we illustrate this need with an example user story and associated scenario to highlight the problems of the current document-centric information flows and how our proposal SKG4EOSC innovatively addresses them.

User Story

As a researcher, I want to discover relevant work in a research area to get an overview of the state of the art.

Scenario

Catherine is beginning her doctoral studies with a focus on circular economy. As a newcomer to this research area, she uses digital libraries of major publishers to discover relevant work. In this way, she learns which approaches are currently leading and how they are evaluated. Catherine will skim through hundreds of papers, most of which turn out to be irrelevant for her research. Dozens of articles will need to be read in detail and the essential information Catherine needs for her research will be manually organized.

Problems

• Initially, Catherine does not know the right keywords and mostly finds irrelevant literature.
• Search results include articles rather than the information Catherine needs.
• Because the information in documents is not machine actionable, Catherine spends substantial time manually extracting, organizing, and processing the required information.
• Since Catherine cannot easily share her evolving literature review, Alex - another PhD student - facing the same problems as Catherine cannot build on her (cognitive) work.
• Being a brilliant early career researcher, Catherine will advance the state-of-the-art, but can only communicate her findings with another article.

How SKG4EOSC innovates

• Services for FAIR scholarly information production, curation, and use in EOSC will enable exploiting scholarly information in a fine-grained manner, not merely at the level of articles.
• Researchers are presented with the state-of-the-art information known about a research problem.
• Next-generation semantic publishing tools ensure machine actionability of content at the time of publication.
• Integration of machine actionable content and data pre- and post-publication into knowledge graphs will bring together the legacy and future of our scientific knowledge.

Europe spends ~2.18% of its GDP amounting ~300 Billion Euro annually for R&D*7. In addition, with climate change, de-carbonization, clean mobility, social inclusion and responsibility, supply chain resilience, etc., we face significant societal challenges, which can only be mastered by research and development. With SKG4EOSC we aim to make European research at least 30% more efficient and effective, since the currently extremely cumbersome processes of literature work (search, exploration, and ingestion) will be greatly improved through the SKG4EOSC state-of-the-art overview, querying, analysis, and visualization services. In addition, SKG4EOSC services will help to reduce duplications in research, guide researchers towards important research goals and significantly improve reproducibility and peer-review. Last but not least, the transfer from research to industrial and societal applications will be greatly improved, since industrial and societal stakeholders can easily obtain current overviews of the state-of-the-art about their innovation challenges.

Table 7 explains how SKG4EOSC applies the FAIR data principles. Table 8 summarizes the steps towards the achievement of the expected impacts of the project over time, including beyond the duration of the project. Table 9 summarizes the SKG4EOSC’s unique contributions towards the outcomes specified in the topic of this call and the wider impacts specified in the respective destination.

Efficient, machine supported, use of scholarly information has been, is and will be a need for research. Standing on the shoulders of giants relies on literature reviews; each community periodically conducts systematic reviews; synthesis of published results is performed to increase the statistical power of claims. All these research activities rely on information extraction and organization from literature, processes that are currently manual and inefficient. SKG4EOSC is the first EOSC project that tackles this problem head on. In the first phase, SKG4EOSC develops its service offering for FAIR scholarly information production, curation and use in the EOSC for four communities directly involved in the project. In a second phase, 1-2 communities will be additionally onboarded during the project as further pilots not directly involved in the project developments. In parallel, ORKG has been onboarding diverse communities through ORKG Observatories. Overall, SKG4EOSC thus reaches tens of thousands researchers during its project lifetime, a reach that is a key pathway to ensuring the project's impact.

By leveraging approaches and enabling technologies for FAIR scholarly information production in both pre-publication and post-publication phases, SKG4EOSC not only develops services that extract information from millions of existing legacy documents and the millions that are going to be written in the coming years but, importantly, also develops a pathway for a future in which scholarly information is produced FAIR during the research lifecycle. While this approach comes with its own challenges (e.g. considerable upgrade of the research infrastructures and tools currently in use) the pre-publication approach (WP2) has the potential to fundamentally transform the production of FAIR scholarly information.

Beyond the project's lifetime, the SKG4EOSC service offering will reach further communities. This is ensured by sustained operations of the services, which is guaranteed by our infrastructure partners. ORKG as the proposed Hub and the Hub-enabled services are and will be sustainably operated by TIB. Disciplinary scholarly information infrastructures operated by partners (e.g. IGB, Wikimedia, etc.) have their own sustainability plans. In the event that a service is retired, we will leverage the SKG4EOSC approach for Nanopublication-based decentralized harvesting to ensure the respective content continues to be available through ORKG. TIB will thus actively coordinate with partnering infrastructures to ensure content and service availability beyond the project's lifetime. Through these measures, SKG4EOSC and its service offering has the potential to reach a majority of researchers and their respective research communities to ultimately fundamentally transform scholarly communication from being purely human actionable to being also machine actionable, eventually reaching the envisioned 30% efficiency gains in annual R&D expenditures in Europe, and globally. Hence, SKG4EOSC lays the foundations to fundamentally transform the way researchers create, share and exploit scholarly information, as well as the way the public and private sectors can exploit scholarly information. Additionally, SKG4EOSC will sustainably address an important gap in the EOSC, namely the seamless access to and management of increasing volumes of scholarly literature following the FAIR principles. Finally, with improved machine actionability in scholarly communication and consequent service offering for machine supported processing of FAIR scholarly information, including the reliable tracking of its provenance and the structured description with formal semantics of materials and methods, SKG4EOSC will substantially contribute to improving the reproducibility of and the trust in science.

The desired global impact of the SKG4EOSC service offering relies on a concerted effort to lift FAIR scholarly information to a first class citizen status in the ecosystem of research objects. As the primary artefact in scholarly communication, articles have held this status for centuries. Required is the same for a corresponding machine actionable expression of scholarly information. The SKG4EOSC project will pave the way, but more investment and regulatory actions will be needed for a global transition. As it can be automated only to a certain extent, the production of FAIR scholarly information relies on researchers and will need to be incentivised, primarily through excellent services that directly add value to researchers.

The desired global impact can face a number of potential barriers. First, change may be very slow, primarily because research and research practices are ingrained activities with established methods and tools that are difficult to advance and steer in new directions. Secondarily, change may be actively resisted by actors that perceive FAIR and open scholarly information as a threat, e.g. to their business models. Furthermore, technology may not mature as fast as needed for the problem at hand. Of particular concern are text and data mining as well as natural language processing, and thus our inability to efficiently extract granular information from documents. This area of research has a decade old history and has not yet achieved the performance needed for scholarly information. Hence, the technological maturity in this area may be a potential barrier, especially in scaling the SKG4EOSC service offering to the massive corpus of legacy articles. An additional potential barrier may be that crowdsourcing does not perform in the scholarly context for scholarly information as well as in other contexts, e.g. for encyclopedic or geospatial information. Successful crowdsourcing typically relies on the 90-9-1 rule, whereby 90% of users only consume content, 9% of users curate existing content and only 1% of users create new content. It is unclear whether in the scholarly context, we can rely on a mere 1% of researchers to produce FAIR scholarly information. Moreover, advancing the existing research infrastructure so that scholarly information is produced FAIR may also prove to be a mammoth endeavour. Hence, the pre-publication approach proposed by SKG4EOSC will also come with significant barriers. Finally, the regulatory framework may not give enough emphasis on machine actionability of scholarly information in the context of Open Science, the EOSC and equivalent international initiatives.

Naturally, progress will occur, in technical as well as in human infrastructures. Computer science will make further progress, especially also on information extraction, meaning that our ability to extract granular scholarly information from the literature is likely to further improve. Modifying the existing and future research infrastructures, services and tools, both open source and commercial so that scholarly information is produced FAIR at birth, is less of a technical challenge. Indeed, the technologies required to do so exist and are mature enough to be adopted in production environments. Along this dimension, the evolution needed is in social infrastructures, especially the willingness of researchers to adopt more advanced services and tools as well as the willingness of commercial toolmakers to advance their systems. As our understanding of how the use of FAIR scholarly information benefits stakeholders will evolve, it will become clearer how these aspects can be accordingly incentivised.

Given that all approaches - post-publication text and data mining, pre-publication FAIRification, and crowdsourcing - have their potential barriers, SKG4EOSC builds on all three approaches to FAIR scholarly information production and curation in order to mitigate the respective barriers. SKG4EOSC argues that the challenge relies on the effective integration of automated and manual approaches.

Dissemination. SKG4EOSC will tailor various uni- and bi-directional dissemination channels to the needs of each defined potential user group and audience, eliciting expertise, knowledge and perceptions from stakeholders as part of the project’s co-design engagement activities. The preliminary mapping of dissemination channels, target groups, related impacts and relevant KPIs for which they will be applied can be seen in Table 11. All SKG4EOSC partners will be actively engaged in the dissemination process by:

• Providing content to the Communication and Dissemination work package;
• Using own personal and/or institutional networks, social media and websites to promote the project;
• Using relevant conferences to present the project results and distribute dissemination materials;
• Publishing research and data papers in reputable international scientific journals, in line with their academic and institutional policies;
• Participating in campaigns and events (conferences, expert round tables, webinars, and workshops) specifically designed to raise visibility of the new community and increase engagement from relevant actors beyond the project consortium.

These efforts will be streamlined in the project Plan for Exploitation, Dissemination and Communication (PEDCOM), which will be a regularly updated ‘living’ document, serving as a management tool for dissemination actions, available to all partners from M6 and updated in M24. These updates will include any necessary modification and adapt appropriately to project progress and new circumstances, including feedback from stakeholders and end-users.

Exploitation. To maximise the exposure of project results and their potential for exploitation, the project will take advantage of the EC’s Horizon Results Platform and appoint Pensoft as a lead partner for these activities. This platform will serve as a bridge towards policy-makers and researchers, giving access to the project's main and prioritised results with a high potential value (Key Exploitable Results, Table 10). In addition, SKG4EOSC will consider the Horizon Results Booster for dissemination and exploitation of results so that the added value of the KER is amplified. The exploitation and sustainability of the SKG4EOSC results and products assumes two levels of responsibilities: (1) products and services developed at the base of either project partners or RI will be a responsibility of the respective partner or RI; (2) the key synthetic product of SKG4EOSC, namely the ORKG Hub, will be hosted and run after the project end by TIB at the first level and a consortium of projects partners and RIs who provide services through the Hub at the second level.

The sustainability will be enforced by the uptake of the products and services by the starting community through actions and measures described in the PEDCOM (D6.2). An essential element of the project sustainability is the adherence to the long term data preservation and accessibility via the repositories and RIs involved in compliance with the EOSC long term sustainability plans supported by the Member States and infrastructures. To ensure also the long term commitment to Open and FAIR data, SKG4EOSC will adopt whenever relevant the RDA FAIR Maturity KPIs to check the Fairness of the data infrastructures involved*8.

Communication. In order to achieve maximum exposure and impact, we will prioritise our communication channels based on the ones that are actively used by our target audiences. With our professional audiences our focus will be on establishing two-way communication, whereas for the general public we will adopt the “getting our message out there” mantra (mainly through collaboration with associations and networks, press releases and social and mass media).

Internal communication. The SKG4EOSC website platform will consist of a password protected internal communication platform (ICP). The ICP will have the following main features: internal repository where all registered users can upload files and all internal documents related to the activities of the project will be stored; a user section containing the profiles of all project members that are granted access to the ICP; upload options files with restricted access, intended only for the consortium members; option to upload news and events; dissemination report forms (symposia & meetings, general dissemination, scientific publications and open research data); living documents containing a view-only copy of important project forms and tables (including contact lists and dissemination reports); a comprehensible step-wise ICP user manual. Additionally, a business communication GDPR-compliant platform will be implemented as a central channel for internal communication. The platform allows for an easy exchange of messages and calls, hence avoiding the need of unnecessary email exchanges.

External communication. SKG4EOSC external communication strategies will be bi-directional, i.e. not only disseminating project outputs to targeted actor groups and the public at large, but also eliciting expertise, knowledge and perceptions as part of the project’s engagement activities. A short guidebook with standard processes and best communication practices (a Handbook of Communication, MS6.2) will be created, together with the communication strategy document. This document will include issues like:

• tips and tricks on how to create social media posts, news articles, press releases and policy briefs,
• instructions on how to shoot and create engaging videos,
• guidelines on how to acknowledge EU funding etc.

The different Communication & dissemination tools and targets, measures to maximise impact and KPIs are outlined in Table 11.

Open data and open science strategy. The Horizon Europe work program highlights the need to have research data and software tools openly used, by maximizing open science practices, access and re-use of all research cycle outcomes. To coordinate the research data management within the project, SKG4EOSC will develop a guiding Data Management Plan (DMP) (D7.2). The DMP will specifically cover: handling of research data during and after the project; data collection and processing; methodologies and standards; data sharing and open access; curation and preservation. The DMP will also provide the dataset metadata specification that will be used in the data registry, following an appropriate relevant standard. It will specify the recommended licensing schemes, preferably using the Creative Commons Public Domain (CC0) and Attribution (CC BY) licenses as suggested by Horizon Europe. In the cases where the datasets cannot be publicly shared, the reasons will be mentioned in its metadata description (e.g. ethical, rules of personal data, intellectual property, commercial, privacy-related, security-related). Below is a preliminary description of all major points to be covered in detail within the project DMP:

• What types of data will the project generate/collect? Numerous and varied data sets will be collected or generated by SKG4EOSC project partners, including specific data types, e.g. data extracted from literature. The project will not only openly share data, but will provide a unique new level of linking open data between different science domains through advanced Linked Open Data technologies (LOD).
• What standards will be used? To ensure interoperability, the SKG4EOSC project aims to collect and document the data in standardized formats (i.e. RDF or tabular data) to ensure that the datasets can be understood, interpreted and shared with accompanying metadata and documentation and relevant supporting material. Metadata standards will depend on the discipline and/or the methodology that was used to produce the data. SKG4EOSC partners will use both discipline-specific repositories and common/standard metadata requirements and ontologies (example for biodiversity: Darwin CoreTaxPub, OpenBiodiv-O, Ecological Metadata Language (EML), etc.), including generic ISO-90155 compliant metadata libraries dependent on discipline-specific or institutional repositories.
• How will this data be exploited and/or shared/made accessible for verification and re-use? SKG4EOSC data will be openly shared through automated workflows with relevant repositories, both generic (Table 2) or domain-specific (Table 3).
• How will this data be curated and preserved? All data and models, both generated as part of SKG4EOSC and obtained from other sources, will be annotated, using ontology-aligned keywords and meta-tags. Output from SKG4EOSC will be organised in an easily accessible and interpretable format. The necessary tools, standards and protocols for making SKG4EOSC data accessible, findable, exchangeable and secured on the long term will be made available to all SKG4EOSC partners and users.
• Management of internal knowledge in SKG4EOSC. The terms of Intellectual Property Rights (IPR) management will be specified in detail in the DESCA Consortium Agreement to be signed at the beginning of the project after discussing and encountering the specific IPR policies and legitimate interests of all partners. SKG4EOSC partners will work on a cooperative basis without commercial interest. However, for future maintenance of software, models and data mutual agreements on ownership and access conditions are essential to build trust and to respect interests relevant for durable cooperation. Issues about ownership, access rights and use conditions will be described transparently in the Consortium Agreement to ensure optimal cooperation among the SKG4EOSC partners. To that end, the Consortium Agreement will define use conditions. User groups, already foreseen in this project, will be asked to agree to these conditions using partnership agreements.
• Management of external knowledge in SKG4EOSC. Non-confidential results will be disseminated on the project website and through open access, i.e. free online access, applying the ‘gold’ open access model. Each WP leader has responsibility to manage external knowledge available to the general public according to the dissemination plan mentioned above. Fundamental scientific results will be freely disseminated through appropriate channels including scientific publications, presentations at international conferences and workshops. The publication venues will be primary scientific high-impact open access journals, especially the Open Research Europe platform. SKG4EOSC will follow the guidelines on open access to scientific publication and research data stated in Horizon Europe. Some budget for supporting publication in open access form will be dedicated and managed by the Executive Committee in accordance with the dissemination plan of the project. Deliverables and other important project outputs will be published in a dedicated open access collection in the open science Research Ideas and Outcomes (RIO) Journal.
• Open source policies. The software tools or plugins produced within the SKG4EOSC will be available as open source code under an OSI-approved license and published in the open science RIO Journal or other appropriate journals to ensure findability and reusability of all open source resources. The aim of SKG4EOSC’s open source approach is to ensure that a framework for new contributions is established that allows them to continue to be developed as open source in order to facilitate the further adoption and update of the technologies by all stakeholders.

The SKG4EOSC work programme is structured in seven WPs (Fig. 6). WP1 establishes the ORKG as a Hub for FAIR scholarly information in the EOSC. WP2 devises innovative approaches for post-publication scholarly information extraction from the literature and related assets, such as figures and tables. WP3 devises innovative approaches for pre-publication production of FAIR scholarly information, especially in data analysis and scholarly communication phases of the research lifecycle. Thus, WP2 and WP3 develop the approaches required in disciplinary scholarly infomation infrastructures to produce the FAIR scholarly information that will be made accessible in a harmonized manner through the Hub. WP4 builds on the Hub and develops Hub-enabled services, as well as their composition with other relevant services in the EOSC. These services build the technical foundations in support of WP5 pilots that leverage FAIR scholarly information and services to support the science underpinning global societal challenges. WP6 ensures to maximize the impact of SKG4EOSC project results through communication, engagement, dissemination, and exploitation instruments and activities. Finally, WP7 is concerned with project management. Each WP has its own roadmap and objectives, and the consortium members will work towards these throughout the project. However, as the high-quality outputs generated by the individual WPs are essential to the success of the overall project, all WPs must strive for success in the overall work plan, requiring close integration and alignment between the executed activities. The time frame of the work plan tasks and milestones is shown in Fig. 7 as a Gantt chart.

Figure 6.  

Work package structure and relations.

Figure 7.  

SKG4EOSC Gantt chart showing Task durations, Deliverable due dates, and Milestones.

The project's objective is to establish an EOSC service ecosystem for FAIR scholarly information production, curation and use by lifting heterogeneous disciplinary scholarly information infrastructures into EOSC and devising innovative services. To match this core objective, the SKG4EOSC consortium involves top leaders in inter-disciplinary technology, infrastructure and service research and innovation and top leaders in disciplinary scholarly knowledge organization. With the ORKG, TIB has established its leadership in the area of FAIR scholarly information with knowledge graph technologies. As SKG4EOSC, TIB brings into the consortium vision and multi-year experience in developing services for FAIR scholarly information, as well as a world-class record in research and innovation on relevant technologies. This expertise is complemented by partner INFAI, which brings into the consortium strong technology knowledge to lead the WP4 developments on ORKG enabled services for FAIR scholarly information use in the EOSC. Technology expertise is further complemented by partner VUA, which has been leading the research and development on Nanopublications, a technology component that will be fundamental to one approach for distributed harvesting pursued within SKG4EOSC. As a representative of Wikimedia and its ecosystem of services including Wikidata, Wikibase, WikiCite, SKG4EOSC partner USFAX brings into the consortium strong technology knowledge, especially in regard to scalable Crowdsourcing. With USFAX, SKG4EOSC will lift the Wikimedia service ecosystem into the EOSC by making services such as Wikidata and Wikibase compatible with the proposed approaches to distributed content harvesting and retrieval. Wikidata and Wikibase will thus be additional service offerings that disciplinary initiatives can employ to manage FAIR scholarly information.

SKG4EOSC builds on numerous and heterogeneous disciplinary scholarly information infrastructures in four disciplines (biodiversity, life sciences, chemical engineering, and social sciences) with their respective partners. With the services Hi Knowledge, Linear Mixed Model KG, and OpenBiodiv, the partners IGB, IPG PAS and PENSOFT have developed state-of-the-art infrastructure and services for FAIR scholarly information production, curation, publishing and use in biodiversity. In SKG4EOSC, these partners will demonstrate the production, curation and use of FAIR scholarly information in a pilot on the biodiversity and ecosystem crisis (T5.1). With the Linear Mixed Model KG, IPG PAS is one of few institutions that have applied the FAIR data principles to scholarly information during the data analysis phase of the research lifecycle. In addition to supporting the biodiversity pilot (T5.1), with its leading experiences in analysis of biological data, IPG PAS will also lead the WP3 developments on pre-publication FAIRification of scholarly information. With Hi Knowledge, IGB has demonstrated disciplinary leadership in information extraction for biodiversity literature, the organization of such information and the development of visualization services using the organized information. With their long-standing experiences in this area and as a strong disciplinary partner, IGB will lead WP5 activities on the development of pilots that will demonstrate the use of FAIR scholarly information in research, building the knowledge base to tackle global societal challenges. With OpenBiodiv and as a world leader in Open Access publishing tools, infrastructure and services, PENSOFT further complements the biodiversity pilot and also brings into the consortium strong technology expertise.

With partners, LUH, SERMAS, UPM, and IDIPHIM, SKG4EOSC will demonstrate the production, curation, and use of FAIR scholarly information in a pilot on precision oncology (T5.2). These institutions have long-standing collaborations in numerous international projects. They merge world-class technology expertise in knowledge graph applications for knowledge organization in the sciences with disciplinary knowledge in life sciences. Partner LUH brings expertise in knowledge graph technologies into the consortium and further complements SKG4EOSC with technical knowledge. UPM are experts in AI technologies for processing clinical data and forecasting patterns for patient stratification. Lastly, SERMAS and IDIPHIM are leading healthcare institutions in cancer treatment and share strong knowledge on cutting-edge clinical methods for cancer patient profiling.

Partners TU Delft and UM are internationally recognized leaders in digital technologies and services for chemical engineering and cheminformatics and will thus demonstrate the production, curation and use of FAIR scholarly information in a corresponding pilot on circular process solutions for the chemical industry (T5.3). As WP2 lead, TU Delft brings into the consortium world class technology expertise in information extraction from literature, text, images and other information assets. Similarly, UM is a recognized leader in developing tools, infrastructure and services for cheminformatics, promoting Open Science, for example in the Journal of Cheminformatics.With the development of Cooperation Databank, partner VUA has demonstrated world class leadership in applying knowledge graph technologies for the production, curation and use of FAIR scholarly information in social sciences. In SKG4EOSC, VUA will leverage their technology expertise and disciplinary knowledge in the development of a pilot on human cooperation and climate change (T5.4).

As such, the SKG4EOSC consortium is a perfectly balanced set of partners, bringing together the necessary disciplinary and inter-disciplinary knowledge to address the project's objective. The consortium includes expertise in social sciences and humanities and develops a corresponding pilot. As underscored in Section 1.2.6, all partners have been leading Open Data, Software, Knowledge and Science champions for decades.

With PENSOFT, SKG4EOSC also involves an SME in the scholarly publishing sector that is well-known among academics worldwide with its technologically advanced peer-reviewed Open Access journals publishing in the domain of biodiversity as well as the development of advanced digital services in scholarly communication such as the ARPHA-XML publishing workflow and the OpenBiodiv knowledge graph participating in this project. PENSOFT’s project department consists of a motivated team of active scientists, project managers and science communicators offering long-standing expertise in delivering the full set of science communication services. PENSOFT has been involved in the science communication of over 30 projects, which guarantees the company’s experience and competence. The company is actively developing new tools, workflows and methods for text- and data publishing, dissemination of scientific information and technologies for semantic enrichment of an articles’ content. PENSOFT will lead WP6 and support communication and dissemination activities for the entire project. PENSOFT will be appointed a lead partner to coordinate the exploitation of results in accordance with the PEDCOM (D6.2, see also Section 2.2). For example, the open source tools developed in the project will be supported by the relevant documentation and made available for use for any interested organisation, mostly industrial entities but also libraries and non-commercial organisations.

Table 12 provides an overview of partner expertise in relevant areas.

The objective of this WP is to harmonize access to FAIR scholarly information in the EOSC by standardising data format and exchange protocols of disciplinary scholarly information infrastructures (directly involved in SKG4EOSC, additionally onboarded during the project, or joining the federation after the project ended). Harmonized access is to enable the efficient development of a wide range of generic and specialized services for the production, curation, and use of FAIR scholarly information in the EOSC.

T1.1 - Hub architecture, (meta)data format and standards, (meta)data exchange protocol

The objective of T1.1 is to evaluate two approaches for harmonizing access to FAIR scholarly information in the EOSC and determine the optimal approach or whether both should be pursued to power WP4 services and WP5 applications. The first approach is decentralized harvesting and leverages Nanopublications. Here, disciplinary infrastructures publish scholarly information as Nanopublications and ORKG harvests the published assertions. The task will evaluate the use of nanopub-servers*10 (Kuhn et al. 2016) for the publishing of Nanopublications by disciplinary scholarly information infrastructures, thus leveraging existing services. The second approach is decentralized retrieval and leverages GraphQL. Here, disciplinary scholarly information infrastructures implement an API of their choice (GraphQL, REST, SPARQL, etc.). Instead of harvesting data, ORKG implements a GraphQL endpoint that enables distributed access to disciplinary scholarly information infrastructures and their content. Both approaches enable harmonized access to FAIR scholarly information and efficient use by Hub-enabled services (WP4).

T1.2 - Hub architecture implementation in disciplinary scholarly information infrastructures

Building on T1.1, the objective of T1.2 is to implement the approaches for harmonizing access to FAIR scholarly information in all disciplinary scholarly information infrastructures. In a first stage, T1.2 will prototypically implement the architectures for the two devised approaches as a minimal viable product. In a second stage, the implementation of the approach that is primarily required by Hub-enabled services (WP4) is then further refined by all disciplinary scholarly information infrastructures for deployment in production environments.

T1.3 - Onboarding additional infrastructures for FAIR scholarly information in the Hub

The objective of T1.3 is to develop the technical specification which current and future infrastructures for FAIR scholarly information are required to implement in order to integrate with the Hub. T1.3 will evaluate the practical viability of the developed specification in onboarding additional (1-2) infrastructures for FAIR scholarly information during the project’s lifetime (e.g. CEUR-WS). The specification will be published Open Access with CC BY license to support future adoption by other infrastructures and thus catalyse the production, curation, and use of FAIR scholarly information in the EOSC.

The objective of this WP is to design and implement open source, web-based interactive tools for the extraction of information from published scholarly articles. This includes the use of natural language processing to extract information from text, computer vision to extract information from images, and the analysis of bibliometric metadata. Moreover, the extracted information is curated to ensure high quality data.

T2.1 - Human-machine extraction of FAIR scholarly information from the literature text

The goal of T2.1 is the design and implementation of an interactive web-based information extraction tool for scholarly literature. This human-centred approach will leverage the joint power of human experts and machine intelligence for understanding and structuring scholarly information at large scale. In the first step, the topics of scientific publications are identified through a (dynamic) topic model that is trained on the abstracts and relevant metadata. In the web-tool, a user selects or uploads a publication within his or her field and marks relevant text passages for information extraction (c.f. building on the ORKG PDF annotator tool*11). The selected text is automatically imported into our tool and a number of pre-processing tasks are executed. This includes the recognition of technical vocabulary from taxonomies. Also, we train and apply a pre-trained BERT transformer model (Devlin et al. 2019) for named entity recognition and relation extraction (Xue et al. 2019). Notably, there exist pre-trained BERT models for multi-domain scientific publications (Beltagy et al. 2019). The results of the named entity recognition are suggested to the user who can confirm or change the model predictions. Similarly, an active learning approach is used to map the entity to a template from a pre-defined ontology. This interactive procedure increases data quality and allows for an active learning approach, which will improve the model performance over time. Moreover, the metadata of the publication, text passage, user ID, and model version will be saved in the KG to improve trustworthiness. This task will initially focus on a specific scientific domain related to the use cases in WP5 with established ontologies (e.g. extraction of chemical compounds and properties from literature). Then, the developed framework will be extended to other domains.

T2.2 - Multi-modal extraction of FAIR scholarly information from figures and tables

Figures and tables contain densely-packed semi-structured information in scientific literature. In T2.2, we implement a service deployable in ORKG or disciplinary infrastructures that extracts information from figures and tables. First, figures and tables will be extracted from the scholarly documents through established tools. We will compare existing tools, integrate the best tool as a service in the ORKG, and improve its performance through an active learning setup. This includes PDF processing toolkits (e.g. PyMuPDF) and computer vision tools (DeepFigures, PDFFigures 2.0, DeepPDF). Second, we aim to extract information from identified figures and tables. While domain-independent methods for information extraction from tables exist (academic (Pinto et al. 2003) and commercial*12), the extraction of information from figures is often domain-specific. We aim to select and integrate an established algorithm for information extraction from tables. For the classification of figures, we will develop a semi-supervised approach that suggests labels based on captions (using named entity recognition from T2.1). A user will check the labels. Then, we will train an image classification algorithm for the labelling of figures. In a subsequent step, domain-specific information extraction approaches will be developed, and existing algorithms will be incorporated. This includes the extraction of information from chemical block flow diagrams using object detection (Kang et al. 2019) and the extraction of chemical structure descriptions using transformer models (c.f. Rajan et al. (2021)). A subset of the extracted figures and associated FAIR data will be uploaded to Wikimedia Commons with structured data annotation for reuse (e.g. by using existing infrastructures*13).

T2.3 - Bibliographic-enhanced information retrieval in scholarly knowledge graphs

There are currently several open bibliographic databases (e.g. DBLP, OpenCitations and Wikidata) that can be used to support the information retrieval of research findings from scholarly publications, including recognition of citation types. Leveraging these resources with rich and interlinked metadata describing the context of research (e.g. publications, datasets, people, organizations, etc.) can bring an added value to the algorithms aiming to extract scholarly outputs from full texts. The objective of T2.3 is to couple these databases with full text analysis (T2.1) and processing using semantic similarity measures, co-word analysis, semantic annotation and advanced machine learning techniques including graph and word embeddings, deep learning and machine learning. This will enable the generation, validation, adjustment and the addition of reference support to FAIR scholarly information published in ORKG.

T2.4 - FAIR scholarly information curation and quality control

A technical challenge in SKG4EOSC is the orchestration of intertwined, iterative data and software lifecycles with the final objective to reach sufficient quality to achieve satisfaction of researcher requirements (end user). Harvested payload (content/data), metadata describing the payload, integration metadata (conversion, links, mappings), crowd- and domain-expert annotations, user-assisting tools, trained and retrained models for these tools are continuously iterated in SKG4EOSC and form a hyper-dimensional Petri net (distributed, discrete event dynamic system) of dependencies that requires efficient, automated and effective quality control. T2.4 will leverage INFAI’s Databus to track and version artefacts, software, metadata in a Persistent ID Graph, that already implements all FAIR principles and will be extended in SKG4EOSC to additionally

1. adhere to TRUST principles (Lin et al. 2020),
2. provide ingrained continuous integration (CI) data quality management processes,
3. measure quality of interconnectedness (links) and
4. deploy pre-packaged “software with data'' developed in WP2 to the EOSC cloud as services.

Building on previous work in the context of DBpedia and on the transparency of Databus, we will implement delegation of data issues in the opposite direction of the data flow from end user over intermediate nodes to source and notify consumers, curators and creators of usage, updates and prioritized issues. INFAI’s role is to develop the Databus platform and collaboratively integrate work by other partners, in particular Trav-SHACL*14 (an SHACL engine by LUH) and other WP2 tasks.

The objective of this WP is to ensure scholarly information is produced FAIR by retaining its semantics and exposing it as FAIR digital objects to scholarly communication (data services and publications). In particular:

• Implement novel technologies and workflows for next-generation semantic publishing of structured content;
• Pre-publication tools for conversion and embedding both human-readable and machine-actionable observational and statistical results into manuscripts and “living articles”;
• Align the use of identifiers and bi-directional linking of FAIRified data between literature and research infrastructures to enable distributed systems to interface with one another;
• Improve and extend automated workflows for seamless post-publication interoperation between knowledge, information and data through RDF-conversion of published content into LOD resources, including Nanopublications and formalization papers.

T3.1 - FAIR scholarly information production in data analysis

The objective of T3.1 is to ensure that scholarly information produced in data analysis is produced FAIR. We will develop methodology to preserve (e.g. statistical) information directly from the computational environments in which analytical results are being generated, and make them machine-actionable for storage and further use in data services and publishing. In addition to disciplinary scholarly information infrastructures, we will explore using the DBpedia Databus platform as a storage for FAIR scholarly information produced in data analysis. Based on initial user requirement analysis, for selected computational environments and data analysis types we will provide tools (e.g. libraries in Python or R) that allow data scientists to document and expose the key results of data processing (i.e., meaningful statistics, such as model parameter estimates or research hypotheses tests (Ćwiek-Kupczyńska et al. 2020)) as FAIR digital objects. The information will be modelled and annotated according to adequate general and domain-specific ontologies (e.g. STATO, QB), integrate persistent identifiers to link contextual entities (e.g. InChI for chemicals, Gene Ontology for biological processes and functions), and made available in EOSC through the ORKG Hub by means of the project-proposed approaches (e.g. Nanopublications). The role of VUA will be to assist with the nanopublication modelling and their integration into data analysis environments.

T3.2 - Novel semantic authoring and publishing tools for structured content

The objective of T3.2 is to develop the ARPHA Writing Tool (AWT) into an independent, innovative, standard-aligned, collaborative authoring and editing environment with extensions and plugins for import of structured data and metadata into texts and semantic enrichment of the narrative. It will integrate narrative and data in an efficient and highly automated way to produce FAIR scholarly information expressed in machine-readable formats (e.g. as ORKG templates or key findings expressed in Nanopublications) before the manuscript is submitted to journal publishing workflows, including conversion to Linked Open Data. We will thus take a definitive step towards resolving the “PDF impediment” to knowledge sharing through facilitating data-driven and semantically enriched publishing, based on both generic and domain-specific standards and ontologies. The AWT will generate JATS XML documents and include persistent identifiers (e.g. DOI, compact identifiers, etc.). These augmented documents can be used for:

1. submission to journal publishing workflows, for example the TIB Open Publishing or WikiJournals, or
2. computational workflows for efficient harvesting and further re-use of semantically enriched content.

SKG4EOSC will demonstrate the workflow through publication of its entire research cycle in a project-branded collection in RIO (T6.3). The role of VUA will be to assist with the nanopublication modelling and their integration into the writing tool and the final publications.

T3.3 - Integration of FAIR scholarly information from research assets

Research assets are structured and information-rich data files that are critical for the reusability, reproducibility, and thus credibility of research results. Common research assets that are published together with scholarly publications include computer code, models, simulation files, electronic lab notebooks, data management plans, supplementary media files, and workflows. The automated integration of information from those assets with the publication itself is highly desirable. The objective of T3.3 is to develop information extraction services for research assets. This includes the use of specialized NLP tools (e.g. FALCON) for named entity recognition of text-based files (e.g. code). Domain-specific tools for information extraction from research assets are also used and further advanced. This includes information extraction from chemical process simulation files, where adoption of interoperability standards will be coordinated with the GO FAIR Chemistry IN. Extracted scholarly information is represented as Nanopublications using suitable templates. The role of VUA will be to assist the nanopublication modelling and to create the needed semantic templates.

The ultimate objective of this WP is to develop novel services for FAIR scholarly information and enable use of the services in the EOSC platform. The developed services will facilitate the work in WP5 in relation to the identified global challenges. WP4 will ensure that the developed approaches in WP2 and WP3 are generic enough to be utilized in novel, yet unknown scenarios. Also, the goal is to ensure WP2 and WP3 approaches are composable services.

T4.1 - State-of-the-art (SOTA) comparisons for research questions

The objective of T4.1 is to enable retrieval of state-of-the-art information for concrete research questions. This will be implemented as a service where users can explicitly state (a) research question(s) and retrieve a state-of-the-art comparison for the provided question(s). The service will be built on top of the foundations of the ORKG comparison capabilities. A critical, non-obvious objective of T4.1 is data quality (DQ) of created SOTA objects in particular regarding breadth and coverage to achieve complete, unbiased (and potentially inter-disciplinary) SOTA comparisons. We will leverage T2.1-2.3 and implement use of ontologies for research questions (IGB).

T4.2 - Intuitive visualisation and exploration of scholarly information

The objective of T4.2 is to develop customizable services for visualisation and exploration of FAIR scholarly information published by ORKG. This will make the access and exploitation of the offered information more efficient. The service will provide faceted search capabilities and personalized retrieval of scholarly information. The service will support scientists to get an overview and latest advancements in a particular field via an interactive UI. A particular focus will be put on the temporal aspects, e.g. how cooperation research changed over time (Balliet et al. 2020). The service will provide predefined visualisations, which will be designed in close collaboration with the domain experts in WP5. Visualisations will be provided in a format which will be easy to use and integrate in scientific papers, with features such as citable snapshots to enhance reproducibility.

T4.3 - Research question answering

The objective of T4.3 is to develop different information retrieval services utilizing question answering (QA) methods. Building on the extracted data resulting from T2.1 and T2.2, ontology-based machine learning models and natural language processing techniques are employed to expose a new natural language interface for users. The service will enable users to explore research contributions and scientific data via posing queries formulated in natural text and getting precise answers (such as resources, papers, and comparisons). Moreover, with the availability of structured information in the form of a knowledge graph, the services’ QA module will be able to find relations between entities and better aggregate information and deliver answers (Jaradeh et al. 2020). Methods such as named entity recognition and disambiguation, and relation extraction are base techniques for the QA system to build on in order to comprehend a posed question and try to find the answer. For instance, an example question would be “What is the most common machine learning method used by state-of-the-art papers addressing entity linking?”. To answer this question, the schema of the knowledge graph needs to be understood by the system and entities to be recognized. Though extracted research data is stored in a KG, different data objects have different representations in the graph and require different traversals and comprehension techniques. As such, task T4.3 requires different types of QA systems to address different data forms and information representations (e.g. QA on tables, QA on figures, QA on datasets). Such systems will have user interfaces for user consumption and API access for other types of clients.

T4.4 - Discovery and tracking of current research trends

The objective of T4.4 is to develop a service that enables users to discover and track current research trends. The service will exploit machine learning techniques for identification of topics. The service will reuse the models developed in T2.1 for topic identification. Researchers and other interested parties can use the service to track the popularity of particular research topics over time. The service will also enable users to get insight information about user specified topics, but also get information about the currently “hot”, most attractive topics in the recent period. In addition to topic tracking, the service will enable identification and tracking of other scholarly information such as datasets, code and other related materials. The service will be accompanied by an automated notification system, which will in an automated manner announce updates and trends to its subscribers (e.g., RSS feed, Twitter bot, mailing list).

T4.5 - Fine-grained provenance tracking of FAIR scholarly information

The objective of T4.5 is to increase the trust in science by improving transparency and reproducibility of results. Research lifecycles generate considerable amounts of provenance information in each step, and this task will develop an EOSC service for fine-grained provenance tracking of FAIR scholarly information covering the whole lifecycle. The service will relate FAIR scholarly information to primary data, capture and expose contextual information, i.e. the activities and the agents involved in these activities by building on PROV-O. Research results will be made more sustainable via an archiving model that enables to load and re-run archived experiments and analyses on EOSC via the Databus model to guarantee sustainable reproducibility and reusability via composition (T4.6) beyond the lifetime of projects, a big technical challenge in science. A custom graphical provenance browser will support user groups of WP5 in traversing and tracing provenance information.

T4.6 - Composition of services in the EOSC portal

The objective of T4.6 is to enable the composition of the developed and integrated services in the EOSC portal. A common API and protocol will be designed so that the services are easily integrated and composed in novel workflows. The services will rely on common data, which will enable effortless creation of service compositions. For example, using the federated graph visualization and exploration service, a researcher could explore the research knowledge graph and identify a topic associated with a particular research article. Next, the researcher could query the research topic trends service and retrieve the popularity information for this particular topic. The “composability” capability will be exploited in WP5 in domain specific and generic, cross-domain use cases.

This work package involves experts from four domains (biodiversity, biomedicine, chemistry, social sciences) and is a testbed for the functionality and fitness of the services developed in WP1-4. In each domain, a current lack of FAIR information is impeding usage of scholarly knowledge for meeting global challenges. The domain experts will closely interact with their communities as well as with the other SKG4EOSC team members in defining user requirements, in adapting the existing domain-specific services for integration in EOSC, and in testing the services developed in WP1-4. Cross-disciplinary use cases will be constructed to exemplify the operability of the services across domains.

T5.1 - Addressing the biodiversity and ecosystem crisis with FAIR scholarly information

Given that 25% of animals and plant species are threatened with extinction and ecosystems are deteriorating worldwide (IPBES 2019), there is an urgent need for science-based applied solutions. For example, efficient management of invasive species relies on knowledge about their introduction pathways, current occurrences, potential for further spread and impacts. Efficient restoration of ecosystems, as a second example, requires knowledge about methods for facilitating the recovery of soils, re-establishment of species and triggering of ecosystem functions. In the respective research areas, specifically invasion biology and restoration ecology, publication workflows are impeding progress, since much of the locally produced scholarly knowledge remains hidden in PDFs that are hard to discover, not machine readable, and often hidden behind paywalls (Jeschke et al. 2021). Databases exist for providing some of the much needed information, but these focus on particular groups of organisms (e.g. GAVIA on birds (Dyer et al. 2017), GloNAF on plants (van Kleunen et al. 2018)) or on certain aspects of invasive species (e.g. their introduction pathways (Saul et al. 2016) or first records in different countries (Seebens et al. 2017)). For every species management plan or specific restoration project, experts will have to visit the disparate resources bit by bit to retrieve the information they need. A tool that has been developed to address some of these challenges is the hierarchical network of invasion hypotheses at Hi Knowledge. In a current project⁷³, it is being developed into an evolving knowledge resource, structuring and visualizing curated information on research questions, hypotheses and their empirical basis in the domain of invasion biology. OpenBiodiv is another useful research infrastructure providing information on biological species. However, these services are not interoperable, and are known and used only in subgroups of the biodiversity research community. In T5.1, experts in the biodiversity domain will therefore define user requirements and use cases and test which of the two pathways is most useful for enhancing the production, curation and regular use of FAIR scientific data in the field: (a) Nanopublications, (b) feeding the local knowledge into KGs that are then connected via the SKG4EOSC hub. Either way, the services that were previously restricted to a local user group because of limited scope and prominence will become available to the European research community at large and could thus become a central source for scholarly information for the biodiversity domain, providing an overview on current research, easing discoverage of relevant information and allowing close involvement of the research community. All this will bring a steep increase in the usability of scholarly information and thus a much stronger basis for actions against the biodiversity and ecosystem crisis.

T5.2 - FAIR scholarly information for the support of precision oncology

Cancer is a leading cause of death worldwide; it generates a tremendous psychological, financial, and physical burden. Advances in oncological treatments enable the potential control of the disease. However, because of the yearly increase in the medical literature, physicians require enormous hours to maintain track of new medical research, hindering, thus, novel treatments' reproducibility. In this task, FAIR Scholarly Information (FSI) will be created from fine-grained descriptions of oncological literature, clinical trials, and oncological treatment response and disease prognosis of patient populations (e.g. lung and breast cancer patients). FSI will be applied in other populations to evaluate reproducibility. So far, systematic literature review tools (Scott et al. 2021), pre-trained embeddings (Rasmy et al. 2021), and literature mining (Zhao et al. 2020) are devised for solving specific tasks (e.g., disease prediction). Still, considerable manual work is needed to extract clinical variables required for tracing reproducibility of patients’ outcomes and novel treatments. In the context of the EU H2020 project CLARIFY, an oncology KG has been built that comprises biomedical entities and relations extracted from scientific articles and databases (e.g., PubMed and DrugBank), and a fine-grained description of clinical notes for lung and breast cancer patients from SERMAS. Clinical data includes wearables, physical examination, oncological treatments, long-term toxicities, and tests. The CLARIFY KG is used to identify patterns to understand long-term toxicities generated by oncological treatments, and relapse and progression of the cancer. Analytical services against the CLARIFY KG enable the detection of patterns in treatment responses, disease prognosis, and toxicities. However, the lack of FSI hinders reproducibility validation of uncovered patterns and medical outcomes reported in the literature. Based on a specification of user requirements and formulation of use cases, in T5.2 biomedical entities and relations (e.g. genes, proteins, metabolites from biological pathways, and interactions) extracted from the literature will be integrated into an SKG with oncology articles and their comparisons included in ORKG. The SKG will be linked to the CLARIFY KG and existing KGs (e.g. Wikidata, DBpedia, Bio2RDF, and ELIXIR Core Resources). Analytical methods and services for reasoning and decision support will enable real-time systematic reviews, epidemiological studies, and clinical trials based on patient information and related results represented in FAIR scholarly information.

T5.3 - Circular process solutions for chemical industry through linked data

The European Union aims at achieving climate neutrality by 2050. To achieve this goal, the chemical industry requires a transformation towards more sustainable and circular practices, eliminating its dependencies on fossil fuels and limiting its impact on the environment (Clark 2017, Kätelhön et al. 2019). This has further implications for many other fields, like medicine and food production. However, sustainability problems are wicked problems that cover almost every aspect of society and often include unforeseen implications (Norton 2015). Thus, they cannot be solved through traditional disciplinary approaches. One of the main problems is the lack of linked and FAIR multidisciplinary scholarly information that is needed for model building and (optimal) decision making (Weber et al. 2021, Patel et al. 2012). In T5.3, we will extract and link information for the support of holistic (Social-)Life Cycle Assessment of processes and products. This includes a multitude of aspects such as process information from chemical engineering, toxicity information from biology, nutrients in food production, synthesis routes from chemistry, and various regional data (e.g., water availability, human rights, and labour rights). For this, we will identify the key facts in the literature to monitor progress in knowledge around the selected processes and products in the four domains according to specific user requirements and focal use cases that will be developed. For each key fact, essential concepts and identifiers will be defined that will allow information extracted from literature to be analysed and linked to other databases. This will be done in collaboration with European initiatives such as EU NanoSafety Cluster (e.g. H2020 NanoCommons, RiskGONE, SbD4Nano) (Karcher et al. 2018), VHP4Safety, FNSCloud, who are stakeholders interested in this knowledge. Thus, our tool will provide researchers, companies, and policy-makers the linked information for the evidence-based multi-criteria decision-making. The tool supports extraction of product composition, chemical structure and chemical properties or nutritional value data from literature, information about chemical processes, product content, and identification of waste and potential pollution. Extracted facts will be accessible through ORKG (WP1). Finally, molecular property models (e.g., graph neural networks (Schweidtmann et al. 2020)) will be trained on the linked FAIR information and will be used to support the (optimal) identification of circular process solutions.

T5.4 - Living reviews of human cooperation and climate change to support scientists and policy makers

Human behaviour is causing climate change and producing an existential crisis on a global scale. Scientific research about how human behaviour is affecting the environment and climate is moving at a fast pace, including institutional and behavioural interventions (Alló and Loureiro 2014, Nisa et al. 2019, Hornsey et al. 2016). We need to continually update our understanding of this scientific information to build and implement policies that enable humans to cooperate to solve the problem of climate change. Yet, scientific publications are made available in PDFs and datasets and are not easily integrated and meta-analysed. The Cooperation Databank (CoDa) (Spadaro et al. 2020) is a knowledge graph of scientific studies about human cooperation, which can be used to produce on-demand meta-analyses and living meta-analytic reviews. Living meta-analytic reviews are summaries of scientific research that can be automatically updated with the emergence of new scientific findings. The CoDa knowledge graph has a rich description of the provenance of scientific results and has an application that allows users to aggregate scientific results and then analyse how the results vary across different contexts. CoDa, however, currently has no links to environmental or other datasets that allow for living reviews about the interconnection of human cooperation with ongoing environmental changes. Also, CoDa now does not represent different theories and hypotheses, and in T5.4 we will do this by using automated analyses over the text of the papers. Identified theories and predictions about cooperation and climate change could then be tested with on-demand meta-analytic analyses and within living meta-analytic reviews. CoDa is Findable, as it contains rich metadata descriptions, and it is Accessible, as its data is represented in standard format (RDF), and it is Reproducible as it contains clear data usage and access license. That said, CoDa is not yet Interoperable, as it is not fully linked to other datasets. CoDa does not include references to other datasets and is limited in the reuse of vocabularies (both points related to the “Interoperability” principle). Interoperability is needed in order to create living meta-reviews on climate change. In T5.4, we will use the CoDa knowledge graph to build an ontology and knowledge graph of studies about human attitudes, beliefs and behaviours that affect climate change, and the various forms of interventions that have been studied to affect behavioural change. It will then be possible to create queries across the knowledge graph that can create living reviews for scientists and policy makers to keep them up to date with the literature.

T5.5 - Bringing scientific work to the next level with combined SKG4EOSC services

With T5.5, partners from several research domains will demonstrate how the combined application of services developed in WP1-4 can be leveraged to solve challenges scientists are facing across domains. They will also explore ethical aspects of the challenges and solutions, particularly those that apply across domains, such as data sovereignty, knowledge equity or the environmental footprint of knowledge graphs. Recurring challenges are, for instance, the connection of single study results to existing hypotheses and theories (Heger et al. 2020), determination of the level of empirical support for research hypotheses by meta-analyses, identification of research gaps and biases and analysis of the robustness of results (‘reproducibility crisis’)*15 (Spadaro et al. 2020). Respective work on these issues is sparse, disconnected, and occurring in different disciplines. The researchers collaborating on this proposal have made pioneering contributions to these issues, such as the Cooperation Databank providing living reviews of research gaps and biases*16. However, results of such efforts often stay hidden in local research environments or are published in the usual scientific outlets with the known shortcomings. The full benefits of the rich portfolio of services that will be developed in WP1-4 will become exploitable once a combined access via the EOSC is made possible (T4.6). In the second half of the funding period, domain experts will develop use cases to demonstrate how the SKG4EOSC services can be integrated in scientific workflows in order to:

• regularly check for gaps and biases to direct research efforts,
• maximize robustness of results by pointing to areas in demand of repeated analyses, and
• establish procedures facilitating close connection of empirical results with theory and affected communities.

WP6 aims to maximise the impact of SKG4EOSC by communication and engagement with the best practices and knowledge generated by the project with relevant user groups. Using state-of-the-art methods for community engagement, the project has the specific objectives to:

1. raise awareness of the project outputs through a recognisable project brand and active dissemination and promotion;
2. maximise the impact and outreach of SKG4EOSC results to user groups;
3. engage participating research infrastructures in sharing open standards and practices;
4. build a strong scientific community for uptake of results, and
5. maximise access to and re-use of research data generated by SKG4EOSC.

WP6 will be focusing on facilitating access to the knowledge graphs and their backing research infrastructures by modern and technically advanced active dissemination and communication methods, and ensure the sustainability of the project results through engagement with the ORKG Hub developed in WP1. This includes the engagement in EOSC through the deployment of most up-to-date technologies. WP6 will achieve its objectives via an ambitious and structured approach, relying on both best practices in the field of science communication, and continuous monitoring and upgrade of communication and dissemination tools. To reach out to all relevant target groups and share project progress and results, WP6 will implement custom communication and dissemination methods for each group, such as training webinars, hackathons, videos and more.

T6.1 - Establish project’s visual identity, branding, website and promotional collaterals

Within the first months of the project, a recognisable project visual identity will be developed alongside an initial marketing pack (a project logo, brand manual, brochure, poster, stickers, letterhead and presentation templates). These will ensure that SKG4EOSC is communicated effectively and professionally with the objective to raise awareness and build a community from the very start. A modern and user-friendly public website will provide an easy-to-navigate, continuously updated platform, allowing for fast access to general information about SKG4EOSC and its activities. The website will be secured for rapid uptake and long-term persistence by bringing together and multiplying the strong brands of the participating research infrastructures, their extensive experience and existing communities.

T6.2 - Develop SKG4EOSC Plan for Exploitation, Dissemination and Communication (PEDCOM)

A comprehensive Plan for Exploitation, Dissemination and Communication of project results will set the rules and Key Performance Indicators (KPIs) to guide and measure communication activities during the project’s duration. The Plan will guarantee outreach to all target groups and will ensure the uptake of results during and after the project duration by defining:

1. key dissemination actors and targets;
2. key messages that the project wants to deliver;
3. multi-modal mix of communication channels, e.g. social media, events, press releases etc.;
4. bilateral communication approaches, and
5. KPIs for each outreach activity.

The Plan will give special emphasis to engage with stakeholders and to attract new communities of users along the entire data life cycle (see T6.3). A revision of the Plan (M24) will provide a midterm evaluation and upgrade according to the performance of communication and dissemination tools.

T6.3 - Large scale dissemination, community engagement and uptake of project results

These activities will be operating on several levels: internal communication system, based on a professional content management software hosted on EU servers (e.g. Teamwork), open notebook science, open access publication of all important project outcome, blog and public relations interface, social media profiles in Twitter and Facebook, and others, to ensure effective integration, prioritization, cost-effectiveness and sustainability of the community’s communication interface, networking activities and operations during the project lifetime and especially beyond it through the ORKG Hub (WP1). Following a mix of traditional and innovative approaches and best practices in science communication and state-of-the-art tools, and based on the project’s PEDCOM (T6.2), a series of FAIR-by-design dissemination and training tools and events tailored to the needs of the different stakeholder groups will support the knowledge transfer and capacity building (e.g. tutorial video screencasts, demonstration showcases recorded in video, webinars, and other engagement events, see Table 8). In close relationship with the virtual services and use cases developed under WP4 and WP5, a set of 4 expert round tables for tackling user requirements, technology of access, interoperability standards and contributing to WP1 activities will be organized to collate expert advice from various stakeholders, including industrial actors and to foster cross-disciplinary fertilisation (M6.5). The ORKG Hub will further advance the SKG4EOSC approach with virtual workshops, curriculum, best practices and models for collaboration at all scales (Tables 8, 11). To reach a wide, global and cross-disciplinary audience beyond the scientific realm, the project will contribute to the Wikipedia ecosystem, in particular Wikidata and Wikimedia Commons. The most relevant and impactful results will be published in authoritative open access journals and gathered together in a SKG4EOSC-branded collection of articles in the Research Ideas and Outcomes (RIO) journal, together with other community-related documents produced along the entire SKG4EOSC research cycle (grant proposal, methods, data management plan, workshop and project reports and other most important deliverable, standards, guidelines, policy briefs, training programme, etc), thus ensuring the FAIR and Open Science spirit and practice at all instances of the project lifetime.

The aim of this WP is to ensure the high quality level of achievement of the project’s results via the continuous monitoring of the implementation and completion of the project tasks, activities, milestones and deliverables, safeguarding their proper and timely development according to the DoW and the project’s work plan, while ensuring the successful collaboration among the partners.

T7.1 - Agile project management

This task ensures the high quality, efficient and timely administrative coordination of the project. It incorporates Administration Management activities, including procedures and guidelines for activity planning and monitoring, cost and time management, submission of periodic progress reports and cost statements, preparation of annual review reports, review presentations, and timely submission of high quality deliverables to the Commission.

T7.2 - Data and software management plan

Aiming to improve and maximise access to and re-use of research data and software generated by Horizon Europe projects, SKG4EOSC will develop a Data and Software Management Plan within the first six months of the project, publish it and keep it up to date during the project’s lifetime (annual updates). The main aim of the Plan is to adhere to the FAIR data management criteria and thus to leverage openness of the project’s design and results. In particular, the Plan will describe what data/software types, licences, formats, access and archiving will be used within the project. To add value to the Plan, an additional one-pager with Data and Software Management Guidelines will be produced and shared with partners, in order to acquaint them with the recommendations valid for the project and serve as a guiding tool when generating/developing, collecting or using research data or software.

T7.3 - Quality assurance and risk assessment

The task focuses on defining and specifying the appropriate mechanisms and processes that will be established in order to maintain a high quality level in the whole project structure and outcomes. Additionally, T7.3 deals with the identification of potential project management risks and the respective monitoring of each risk profile as well as with the definition and timely application of contingency plans.

T7.4 - Project reporting and communication

The project coordinator will act as the point of contact for partners in communications with the Commission. The coordinator will ensure that the annual reporting to the EC, semi-annual technical internal reporting, milestone review, midterm review will be implemented.

T7.5 - Annual activity reporting

On an annual basis, project reports will be drafted and released focusing on the progress and intermediate results, and updated plans for the following period.

T7.6 - Onboarding the novel services in EOSC Portal

The objective of this task is to ensure that by project end, all adopted and newly developed SKG4EOSC services are discoverable in the EOSC Portal Marketplace (https://marketplace.eosc-portal.eu).