Diving into the ORKG - A Beginner’s Guide
So, you’re new to the ORKG? Welcome and great to have you on board! We understand that getting started with a new tool can be very overwhelming, especially if the tool provides as many features and possibilities as the ORKG does. This introduction will help you to understand the basics of the ORKG and how you can leverage the software for your individual use case.
Drowning in papers - Keep your head above the publication flood
Research is a fundamental pillar of societal progress. Yet, scientific communities face great difficulties in sharing their findings. With approximately 2.5 million newly published scientific articles per year, it is impossible to keep track of all relevant knowledge. Even in small fields, researchers often find themselves drowning in a publication flood, contributing to major scientific crises such as the reproducibility crisis, the deficiency of peer-review and ultimately the loss of knowledge.
The underlying problem is that we never updated our methods of scholarly communication to exploit the possibilities of digitalization. This is where the Open Research Knowledge Graph comes into play!
The ORKG makes scientific knowledge human- and machine-actionable and thus enables completely new ways of machine assistance. This will help researchers find relevant contributions to their field and create state-of-the-art comparisons and reviews. With the ORKG, scientists can explore knowledge in entirely new ways and share results even across different disciplines.
The Open Research Knowledge Graph
As the name suggests, the Open Research Knowledge Graph (ORKG) is a crowd-sourced knowledge graph for scholarly knowledge. We aim at describing research papers in a structured manner to be better able to find and compare papers related to a specific research problem.
The ORKG makes scientific knowledge human- and machine-actionable and thus enables completely new ways of machine assistance. This will help researchers find relevant contributions to their field and create state-of-the-art comparisons and reviews. With the ORKG, scientists can explore knowledge in entirely new ways and share results even across different disciplines.
Features of the ORKG
-
Papers & Contributions: At the center of the current scientific information flow are published documents. With the ORKG, we want to translate the main contributions of scientific papers into machine-actionable knowledge.
-
Comparisons: Comparisons are the core type of ORKG content and give a condensed overview on the state-of-the-art for a particular research question. Contributions towards the problem are organized in a tabular view and can be compared and filtered along different properties. An example from virology could be a comparison of different estimates for COVID-19’s basic reproduction number, in materials science we might be interested in the solubility parameters of different compounds and for computer science, a comparison of algorithms performances can be useful. With ORKG Comparisons, information from dozens or hundreds of papers can be condensed in one overview. If the comparison has numeric results, it can be easily visualized.
-
Reviews: Reviews are a novel ORKG-based method to create review or survey articles for giving an overview on research addressing a particular research question. Reviews are dynamic, community maintained articles, meaning that anyone can contribute to an article (similar to Wikipedia). Articles consist of several ORKG components, including: comparisons, visualizations and individual papers. As the article contents are stored within the graph, the articles are machine-actionable. Other than traditional review articles, ORKG Reviews can easily be updated when the underlying content changes.
-
Lists: With Lists, you can easily collect relevant resources regarding a certain topic. They provide a good starting point for Comparisons and Reviews and can be used to share relevant literature collaboratively.
-
Templates: The structured description of research contributions is no easy task. Describing scholarly information is complicated. You need to decide at what level of granularity you want to describe a research contribution, the addressed problem, its results and employed material and methods. Also, research contributions addressing the same problem should be described in a comparable manner. Finally, while for humans it is best to only capture essential information, for machines shallow structures typically carry little, no, or ambiguous semantics. To address this issue, ORKG supports the possibility of creating templates that specify the structure of content types, and using templates when describing research contributions. Templates make it easier for users to enter information about new papers. Also, describing a sufficient amount of contributions via a template will generate a dataset that can be used to train models for automated information extraction. In the future, the ORKG can then automatically suggest how a template is filled in and the user only needs to check if the suggestions are correct and make minor changes.
-
Benchmarks: Benchmarks and leaderboards allow to organize all contributions towards a clearly defined and measurable benchmark according to various benchmark metrics and are especially relevant in the domain of Computer Science.
ORKG Data
Our goal is to provide the scientific community with high-quality machine-actionable knowledge across all domains. Openness and FAIRness are key values of the ORKG and are an important goal for the entirety of our data.
The ORKG relies on you, the research community, to contribute knowledge from your domain. We often get the question: Why can’t AIs such as LLMs do the job? The reason why we need you as humans to fill the ORKG with valuabel data is simple: Nothing beats the accuracy of a domain expert when describing knowledge of their own domain. While LLMs such as ChatGPT are very convenient, they often tend to hallucinate and are therefore not the best choice when it comes to scientific knowledge where we want the highest level of preciseness. If you are now thinking that this requires a lot of effort from you, you should read about all the benefits of contributing to the ORKG.
However, AI tools can be a useful support in most ORKG workflows. If you want to know more about how to leverage ChatGPT to create ORKG comparisons, you can take a look here. Furthermore, we also have inbuilt AI support, that can aid you in finding the right properties and resources based on the paper abstract given that the paper abstract is provided. You can find more information about that here.
First steps
You have now read all necessary background knowledge about motivation and the rough idea behind the ORKG. Now it is time to become active and dive into one of ORKG’s many features available at the ORKG main page. If you want to do that alongside the ORKG Academy, you can start by following along one of these courses: