According to the UNESCO Institute of Statistics, we are spending almost US$1.7 trillion per year worldwide for acquiring new knowledge through research (UNESCO Institute of Statistics 2020). Currently, however, this is not a good investment and, each year, an ever-increasing share of this investment is wasted. The reason for this is that, for representing and sharing research findings, we use antique methods, which were developed many centuries ago. Since the beginning of modern science – with the publishing of the first scientific journals – the Journal des Sçavans and the Transactions of the Royal Philosophical Society in 1665 (Mack 2015, Spinak and Packer 2015) – we use the same methods for representing and sharing scholarly knowledge: scientific articles. At the time of the polymath Gottfried Wilhelm Leibniz in the 17th and 18th centuries, a single researcher could still read the entire published scientific literature.

Today, each year, 2.5 million new research articles are produced. Even in a relatively narrow scientific field, it is impossible to read, comprehend and make sense of all scientific articles. For example, publications from 1980 to 2012 show an exponential growth rate of 3% annually (Bornmann and Mutz 2015).

For the genome editing method CRISPR/Cas9, for example, the research search engine Google Scholar lists a quarter-million publications available as PDF articles. If a researcher is interested in how good the method is compared to other genome editing methods, what specifics it has when applied to insects and who has applied it to butterflies, a researcher needs either years of experience or is very likely not to find what he or she is looking for. Imagine that, to order a new iPhone, you had to compare prices by checking dozens of mail order catalogues published as PDF or, to navigate to a hotel, you would need to look at a PDF scan of a street map. This is exactly how the exchange of research findings works today – the previously analogue articles from scientific journals are now made available and distributed as PDF documents.

The new methods of the digital world, such as filtering large amounts of data and information, integrating information from different sources or involving users via crowdsourcing to review and help to organise the information, are non-existent in scholarly communication. Researchers are drowning in a flood of millions of pseudo-digitalised PDF publications. As a result, some research is seriously flawed: many research results cannot be reproduced by other researchers, peer-review struggles to cope with volume, speed and quality and we have more and more redundancy. Major social challenges, such as handling the COVID-19 pandemic and infodemic (WHO 2020) or implementing climate neutrality, require interdisciplinarity and putting bits and pieces from different disciplines together, which is currently extremely cumbersome and resource-intensive.

In the ERC ScienceGRAPH project, we are researching and devising foundational concepts for organising scholarly communication in a knowledge-based way, leveraging a new formal model – cognitive knowledge graphs. According to this model, research contributions are represented in a human and machine-readable manner – the knowledge graph. As a result, completely new ways of machine assistance, such as semi-automatic generation of state-of-the-art overviews, visualisations or even question-answering applications become possible. To prepare the demonstration of the ScienceGRAPH results, the ERC project partner TIB Leibniz Information Center for Science and Technology (also directed by the ERC grant holder Sören Auer) started to develop the Open Research Knowledge Graph service, available at https://orkg.org. As an example, Fig. 1 shows a state-of-the-art comparison of different studies targeting the research question about the R₀ base infection rate of COVID-19.

Figure 1.  

State-of-the-art comparison of different studies targeting the research question about the R₀ base infection rate of COVID-19.

Based on such a structured semantic and machine-readable representation, various other exploration and assistance tools are also possible, for example, a chart visualisation, aggregating the results from the various studies. This example illustrates the solution to problems for various stakeholders:

• Researchers in the field (here epidemiologists and virologists) can get a quick overview of the state of the scholarly discourse related to a particular research question and determine gaps or how they can devise their approach to make their contributions stronger.

• Peer-reviewers can quickly assess the merits of a particular approach and view it in comparison to the current state-of-the-art.

• Publishers have a tool for assisting their editors, editorial managers, reviewers and authors to make contributions stronger and better positioned in the scientific discourse. In addition, publishers using such semantic descriptions and comparisons will dramatically increase the attraction of their journals.

• Equipment and instrumentation manufacturers can ensure that important configurations of materials used in research are documented and the use of their devices is properly acknowledged and visible.

• Industrial and societal stakeholders get faster and better access to the state-of-the-art and can, thus, more efficiently and effectively realise research-based products and services.

While some user groups will not pay directly for this solution (e.g. researchers and peer-reviewers) and the ORKG will be an open infrastructure in general, we see the potential for various value-added services for publishers, equipment and instrumentation manufacturers and other industrial and societal stakeholders.

To realise the potential, with this ERC PoC project, we aim to demonstrate some key results attained in the first two years within the ORKG.org proof-of-concept:

• Integrate the crowd- and expert-sourcing authoring and curation model for cognitive knowledge graphs, based on the knowledge graph cells concept (Vogt et al. 2020).

• Integrate persistent identifiers for scientific sensors and instruments to support the provenance and reproducibility of research results from experiment to publication.

• Develop approaches for generating comprehensive state-of-the-art overviews for a specific research question from the semantic knowledge graph representations of corresponding contributions.

The ORKG is completely unique in its idea to describe scientific contributions in a knowledge graph. There are several other knowledge graph projects for scholarly communication also from commercial players, such as SciGraph from Springer Nature or the Microsoft Academic Graph. However, these initiatives solely focus on bibliographic information and do not comprise a rich structured representation of the actual content of the publications. Other related initiatives are text-mining projects, such as SemanticScholar, which generate some relatively shallow semantic descriptions automatically. However, due to the low precision and recall of text mining methods (in particular for relation extraction), this does not go beyond relatively simple classifications, annotations and summarisation of the content and, thus, does not suffice for creating a comprehensive knowledge graph representation and exploration services, such as comparisons, visualisations, question answering, etc.

To maximise the societal benefit from the results of the ERC and this PoC project, the core ORKG service will be an open infrastructure following the Open Science, Open Access and Open Source principles. This also enables rigorous and large-scale testing and evaluation of the outcomes of the project with real user communities. TIB is prepared to sponsor and further develop, maintain and operate the ORKG service in the long term. In addition to the open strategy, we envision various commercialisation opportunities including:

• Providing value-added services tailored for commercial scientific publishers, such as Springer Nature, Wiley and IEEE Publishing.

• Providing commercial data, analytics and question answering services for speeding up the spread and transfer of research results in industrial applications.

• Partnering with industrial stakeholders, in particular scientific instrument manufacturers, regarding sponsoring of the ORKG and integration of their instrument descriptions.

TIB has long-term established R&D collaborations and customer relationships with small and large industrial stakeholders. TIB already provides commercial literature access services to > 100 customers and aims to expand this to the research analytics services offered on the ORKG service infrastructure.

Since the ORKG project is relatively focused, we envision a lean organisational structure depicted in Fig. 2.

Figure 2.  

Organisational structure and decision-making process.

In addition to the PI and the ORKG development lead, the organisational structure will involve leads of the three ORKG work packages, an industrial advisory board as well as an ORKG community board.

The industrial advisory board will advise the project team in matters related to the commercialisation of the results, such as product features, product and service offerings, IPR, pricing, as well as legal matters. We will organise quarterly meetings of the board. We have been in touch with several industry representatives about joining the board.

The ORKG Community board will advise the development team with regard to community requirements and comprise experts from various research fields, research data infrastructures and open-access publishers. We plan to organise quarterly webinars or workshops with the community advisory board (possibly in conjunction with larger scholarly communication events).

Decision-making and development methodology. The size of the project allows it to follow a lean focused decision-making process, where most of the decisions are made in the regular weekly ORKG project meeting by involving the whole team. For all developments, we follow the agile KANBAN-inspired development methodology aiming at establishing a constant active development process by optimising the issue burn rate and establishing a proactive communication culture.

We aim at identifying, evaluating and eliminating or minimising potential risks that may jeopardise the success of the project. While some relevant project risks and how to address them are already identified, risk management will be conducted throughout the project. It is a continuous process in which known risks will be regularly reviewed and new risks will need to be recognised to handle and control them adequately. Their assessment will lead to the formulation of appropriate mitigation measures that should help to prevent and overcome a risk or reduce its effects to an acceptable level. The process behind risk management can be broken down as follows:

1. Risk identification (i.e. recognise and describe risks).

2. Risk analysis (i.e. analyse likelihood and consequences of risks).

3. Risk assessment (i.e. determine magnitude/acceptability of risks for the project).

4. Risk response planning (i.e. create and execute an action plan to prevent or minimise risks).

5. Risk control (i.e. monitor, track and review risks and mitigation actions).

Table 1 contains some examples of risks and corresponding mitigation strategies that we already identified.

The team is led by ScienceGRAPH PI Prof. Dr. Sören Auer. He is supported by the ORKG project head Dr. Markus Stocker, who has been leading related research and development activities for almost two years. In addition, a seasoned team is already established, including experienced PostDoc researchers (e.g. Dr. Jennifer D’Souza and Dr. Lars Vogt), more than five PhD students, software developers (Manuel Prinz and Kheir Eddine Farfar) and business and technology transfer experts (especially in the TIB departments), which can be dynamically involved in the project as required.

Sören Auer. Following positions at the Universities of Dresden, Ekaterinburg, Leipzig, Pennsylvania, Bonn and the Fraunhofer Society, Prof. Auer was appointed Professor of Data Science and Digital Libraries at Leibniz Universität Hanover and Director of the TIB in 2017. Prof. Auer has made important contributions to semantic technologies, knowledge engineering and information systems. He is the author (resp. co-author) of over 200 peer-reviewed scientific publications. He has received several awards, including an ERC Consolidator Grant from the European Research Council, a SWSA ten-year award, the ESWC 7-year Best Paper Award and the OpenCourseware Innovation Award. He has led several large collaborative research projects, such as the EU H2020 flagship project BigDataEurope. He is co-founder of high potential research and community projects, such as the Wikipedia semantification project DBpedia, the OpenCourseWare authoring platform SlideWiki.org and the innovative technology start-up, eccenca.com (now employing more than 40 people). Prof. Auer was founding director of the Big Data Value Association and led the semantic data representation in the International Data Space.

Dr. Markus Stocker is head of the Knowledge Infrastructures research group at TIB. Markus holds a PhD in Environmental Informatics from the University of Eastern Finland, an M.Sc. in Environmental Science from the University of Eastern Finland and a Diploma (M.Sc.) in Informatics from the University of Zurich, Switzerland. He is author of 40 peer-reviewed journal and conference proceedings papers, with more than 1000 citations. He has managed partner contributions and been involved in various H2020 projects, including THOR, ENVRIplus, OpenAIRE, FREYA, ENVRI-FAIR, as well as nationally-funded projects in Finland and Germany. Prior to TIB, Markus held a postdoctoral research associate position at PANGAEA, the Data Publisher for Earth & Environmental Science, at the MARUM Center for Marine Environmental Sciences, University of Bremen, Germany. As a member of the Research Data Alliance (RDA), Markus is involved in various groups, in particular the WG Persistent Identification of Instruments and the IG From Observational Data to Information. He has several years of professional experience in software development and semantic technologies, with positions at Hewlett Packard Labs, Bristol, UK and Clark & Parsia, Washington DC, USA.

Alexandra Garatzogianni is the Head of the Knowledge and Technology Transfer Department at TIB, leading a diverse and inclusive team, which, besides providing impulses for innovation and developing future technologies, offers comprehensive consulting, research and support in order to enable sustainable access to the market for research output. She is the Coordinator of the H2020 project TRUSTS Trusted Secure Data Sharing Space, of the H2020 MediaFutures, Data-driven innovation hub for the media value chain and WP leader for the H2020 project PLATOON (Digital PLAtform and analytics TOOls for eNergy). She leads the project management of the Leibniz Joint Lab Data Science & Open Knowledge amongst TIB, the Leibniz University of Hanover (LUH) and the L3S Research Center, which serves as a nucleus for further initiatives in the field of research and innovation. She co-founded the IDSA Competence Center at the Leibniz Joint Lab Data Science & Open Knowledge in June 2019 and received the BDVA iSpaces award on behalf of the L3S Research Center (November 2019), which signifies that L3S is a Trusted Data Incubator targeted to accelerate take-up of data-driven innovation in commercial sectors.

The role of the PI Prof. Dr. Sören Auer is to develop and communicate the strategic vision of the project and to devise the key development milestones and priorities. He will advise and mentor the PhD students and PostDocs on the project and work closely with the ORKG development lead Dr. Markus Stocker. A further focus of the PI is to build strategic partnerships, attract further funding, sponsoring or investments. The ORKG project development head, Dr. Markus Stocker, will lead the day-to-day operations and developments of the project. He will lead the regular KANBAN sessions together with the development deputy Manuel Prinz and guide the research and development along with the community and advisory board defined requirements and strategic priorities. Alexandra Garatzogianni will lead the business development strategy and contribute to building and maintaining sustainable sponsor, partner and customer relationships for the ORKG service ecosystem throughout and beyond the project’s duration. The main strengths and weaknesses of the team include the following:

Key strengths

• Successful track record of translating research excellence into large scale applications including successful commercialisation in a spin-off.

• A long history of industrial collaborations.

• ORKG innovation concept with an enormous value potential.

• A seasoned team including a variety of backgrounds and skills: experienced PostDoc researchers, PhD students, software developers and business experts, who can be dynamically involved in the project.

Weaknesses

• Limited resources compared to commercial entities (e.g. commercial publishers).

• Community and industrial buy-in just starting to develop.

• More advocacy, policy backing for the transition/digitisation in scholarly communication required.

• Initially limited possibilities for automation using AI and machine learning due to the lack of training data.