An Open Research Knowledge Graph is a knowledge graph that contains, in particular, scholarly information that is published and communicated in scientific literature – in such a way that the information is machine readable. Machine readability is the key aspect of knowledge graphs. After all, the current method of communicating scholarly information – using natural language, data in tables and images as digital PDF documents – limits the extent to which machines can help us to search and explore information.
Try searching Google Scholar for scientific articles published on highly significant statistical hypothesis tests involving certain variables and sample sizes. Google Scholar and other search engines are unable to answer such queries with sufficient accuracy because the necessary information base is not sufficiently accessible to machines. The ORKG seeks to give machines better access to a wider body of scientific information, opening up new possibilities and setting new standards for scholarly communication in the 21st century.
A revolution in science – this claim is often heard in connection with knowledge graphs. How exactly can knowledge graphs change access to knowledge?
The way in which scientific findings are communicated has remained virtually unchanged for centuries. It must be said that documents are now available digitally, and information systems facilitate full-text search. And yet, in the age of modern information infrastructures, it is unsatisfactory to continue presenting scholarly information solely as text-based documents containing graphic representations of tables and images.
Of course it is often a pleasure for us to read a well-written article, sipping a nice glass of red wine. The possibility to continue doing so should – and will – remain. It is important, however, that more content should become more easily accessible to machines, making it easier for us to find the well-written articles and to process the information they contain more effectively. Not only do knowledge graphs support more precise searches, they also enable the content of scientific literature that describes the problem, the methods and materials used, or key results to be linked flexibly, enhancing knowledge in the process.
What is also interesting is the extent to which such information systems simplify the exploration and presentation of the state of the art, and how the state of the art of a particular issue or problem has developed over the decades. The main objective is to take scholarly communication into the 21st century by presenting scientific information in a structured and machine-readable format, making it more accessible and interpretable to people and machines alike. As such, we can without doubt speak of a “revolution in science”, at least in scholarly communication.
TIB is deeply involved in the topic of knowledge graphs. What is the current state of research, and what exactly are you working on at the moment?
The knowledge graph issue must be viewed and understood in the context of the history and development of the “semantic web” – at least from the perspective of research and development at TIB. This issue, and hence the current state of research, is more or less being driven by the same community, maybe with the exception that, with Google and other large players, principles and, in some cases, technologies are now being used, via knowledge graphs, that were originally embedded in the semantic web. Several Research & Development employees at TIB have built their academic careers on this research area, so it is therefore only natural for TIB to focus intensively on the knowledge graph issue, both in science and in industry. By researching knowledge-based systems in knowledge infrastructures, the “Knowledge Infrastructures” Research Group is making a significant contribution to implementing the ORKG vision.
Knowledge infrastructures are networks of people, artefacts (such as data, software systems and documents) and institutions that acquire, manage and share knowledge. Focusing on science, the junior research group is working on the next generation of artefacts and their ability to help us to capture and manage knowledge. In the process, we embed our work in the research lifecycle, and develop concepts and methods that demonstrate how technical systems will be designed to be more knowledge-based in the future. Semantic technologies and knowledge graphs play a major role in this respect. The research group takes an interdisciplinary approach to explore how such technologies can be integrated into data analysis, scholarly communication, and virtual research environments. In the process, the group addresses information science issues, and collaborates closely with research communities and scientists, particularly in requirement analysis and in embedding technology in real-world applications.