Anna-Lena LorenzNo #EngineeringIntelligence without us!
Women have always played a part in engineering and designing intelligent systems. Starting with early pioneers like Ada Lovelace, the mother of programming languages, over NASA’s Fortran expert Dorothy Vaughan, whose code sent people into space, to the many women working on computer science and AI today, women’s contributions continue to shape the world of interconnected knowledge. However, often these contributions go unnoticed or are attributed to men instead. Time to change that! As we celebrate this year’s International Women in Engineering Day under the theme of #EngineeringIntelligence, let’s take a look at what our female researchers and software developers have accomplished to make the Open Research Knowledge Graph an intelligent system.
Who we are and what we do
The Open Research Knowledge Graph, our lighthouse in the publication flood, is a non-profit platform for machine-actionable scholarly knowledge, combining semantic technologies, AI and human expertise to provide researchers with structured scientific knowledge from publications. Our system helps to find answers to research questions, compare publications, reproduce scientific findings and make research information findable, accessible, interoperable and reusable.
ORKG’s success paved the way for many satellite systems, most notably the TIB Knowledge Loom for truly reproducible and reusable scientific knowledge and the TIB AIssisant, an AI agent supporting scientists throughout the whole research lifecycle.
Many components interconnect to make an ecosystem like this possible, from data ingestion over quality control to displaying results. A team of more than 30 researchers, software developers and curation specialists work together to bring in diverse expertise and realize these components. With roughly one third of our team members, the ORKG team has a relatively high share in female researchers compared to the national average in Germany. In this article, we take a look at the contributions of eight women specifically from our team and their perspectives on the ORKG.
Getting Data In
The knowledge in the ORKG ecosystem has to come from somewhere and while AI extractions can be a good starting point, they do not replace human expertise. Crowd-sourced curation done by domain experts plays a huge role. The challenge is to structure this often unorganized knowledge, especially since domain experts are often not data curation experts. Human-machine collaboration is required. A large part of our PhD student Lena John’s research focuses on Human-In-The-Loop workflows to data curation. With ExtracTable, she developed a tool that creates a scientific corpus of literature, extracts knowledge and compiles it into structured comparisons, while providing user-friendly tabular interfaces for human validation and correction.
Another of her tools, SciMantify turns already existing CSV files into structured knowledge with a guided semantification workflow.
Learning Structure
Dr. Jennifer D’Souza leads an NLP research group at TIB focused on scientific knowledge organization. Together with her team, she develops AI-supported tools and workflows for structuring, aligning, and publishing scientific knowledge. Among these tools are schema-miner, a tool for mining templates from large collections of research papers, making it easier to describe them in a structured, consistent format in the ORKG, as well as OntoAligner and OntoLearner, two libraries for AI-based workflows for ontology alignment across multiple scientific domains and ontology engineering. Together, these tools help make scientific knowledge more aligned and interconnected by using the same vocabulary and schema.
Ensuring Quality
“An application is only as good as the underlying data. As we move toward structured, human- and machine-readable knowledge, ensuring the quality of our knowledge graphs becomes increasingly important.“
Lena John
The SciKGDash Curation Dashboard provides a comprehensive overview of knowledge graph quality through interactive visualizations and supports curation workflows that help identify, track, and resolve quality issues. Its goal is to enable continuous improvement of the ORKG’s data quality and reliability.
Getting Data Out
As our PhD student Golsa Heidari states:
“Good research has little impact if people cannot find it.”
Her work focuses on making research more accessible through intelligent retrieval and knowledge organization. In one of her projects, she built a unit conversion system, allowing harmonization across different studies. By integrating third-party tools for systematic reviews, she makes ORKG even more useful. Additionally, she developed the basis for a dynamic multi-level faceted search for scientific knowledge.
Mutahira Khalid, also a PhD student in the ORKG team, builds on this work with her smart filters, a context-aware filtering service that is proven to effectively reduce search noise.
Making Science Reproducible
“Scientific publishing should not be seen as the final product of research. Rather, it is the beginning of a new cycle where scientific findings contribute to solving complex environmental and social challenges.”
Dr Lauren Snyder
Science lives by exchange and collaboration. One researcher picks up another’s findings, continues their work, proves or disproves their theories, brings in their own perspective or uses old knowledge in a new context. For that, scientific statements need to be reusable, reproducible and open. The TIB Knowledge Loom, co-founded by Dr Lauren Snyder, tackles this challenge. With her background in ecology and biology, Lauren brings her experience with research workflows outside of computer science to enable a seamless integration.
Venturing Into New Projects
Our newest addition to our research projects is the TIB AIssistant, an AI tool that supports researchers throughout the whole scientific process. PhD student Farhana Keya focuses on the very first steps. Her work assists with idea generation, building an environment for researchers to brainstorm based on existing publications. To later on find more material, users can then use her federated research artifact search.
Building A Frontend
As a research web engineer, Qurat-ul Ain Aftab does more than just design a good-looking interface. She engineers the fronted data architectures that make Open Science accessible, bridging the gap between backend systems and seamless user experiences. The template graph view in ORKG is built on her work.
Using The ORKG
Dr Sanju Tiwari earned our curation grant several times by now and used it to prove that ORKG can advance her research on hallucination detection and mitigation in large language models. She already wrote several papers using the ORKG and got a lot of journal acceptances for them. Congratulations!
Why This Matters
Without the work of our many female researchers and developers, the ORKG would not be the system it is today. Just as a machine depends on every component working together, science and innovation thrive when we value and include the contributions of all researchers.
Especially in the age of machine-learning and AI, it is necessary to include diverse perspectives to combat issues like the gender data gap and build systems for everyone.
Only by recognizing and empowering intelligence in all people, we can truly start #EngineeringIntelligence.
For further reading, we reccomend also our blog series Women in Science, featuring interviews with Lena John, Dr Lauren Snyder and many more women at TIB!
#ORKG #Forschung #FrauenAnDerTIB #WomenInScience #LizenzCCBY40INT #ResearchAndDevelopment
Politico.eu (Unofficial RSS)
TechFinitive
Pete Bleackley
Headlines Africa
Analyst207
World news | The Guardian
Felis Floppus