Sie haben Javascript deaktiviert!
Sie haben versucht eine Funktion zu nutzen, die nur mit Javascript möglich ist. Um sämtliche Funktionalitäten unserer Internetseite zu nutzen, aktivieren Sie bitte Javascript in Ihrem Browser.

Glasfasern aus der Optoelektronik in der Fakultät für Elektrotechnik, Informatik und Mathematik, Foto: Universität Paderborn

Show image information

Glasfasern aus der Optoelektronik in der Fakultät für Elektrotechnik, Informatik und Mathematik, Foto: Universität Paderborn

EIM News

| Juuli Eckstein

An interview with Prof. Dr. Axel Ngonga about his award as "Most Influential Scholar"

Prof. Dr. Axel Ngonga, head of the Data Science (Dice) group at the University of Paderborn, is one of the scientists with the highest impact in the field of knowledge engineering worldwide. This was the result of analyses by the independent online platform AMiner, which announced in the recently published “AI 2000 Most Influential Scholar List” the 2000 most influential researchers in the field of artificial intelligence. Prof. Dr. Axel Ngonga, was ranked sixth in the field of knowledge modeling, that records the research over the past 10 years.

In the interview, he talks about his research, which earned him this top position.

What is your research about?

Ultimately, we want people and machines to be able to solve problems together. This means that we try to model knowledge in such a way that it is still understandable for people but that machines can also process this knowledge. Our research is focused on building and using knowledge graphs in a variety of applications. A vivid application example is cancer research, where we model knowledge in such a way that we can process complex inquiries from domain experts in a time-efficient manner. The answers can then be used by doctors, for example, in the personalization of therapies.

What is knowledge engineering?

Knowledge modeling (as it is called in German) in the classical sense is about building models in such a way that data or knowledge-driven systems can answer questions efficiently. The central modeling question for us is: How do we model knowledge so that machines and people can work together?

What exactly have you researched in the past 10 years to achieve such an impact?

A lot of our work revolves around the basics of explainable AI. For example, we investigate how to verbalize AI models and thus explain these models using natural language. In this way, people and machines should not only understand the AI ​​models, but also their formation. This can lead to cooperation between humans and machines by people correcting certain learning steps or rejecting models or making additional training data available. The AI ​​should learn what was actually intended.

Another area is processing requests to large knowledge graphs with billions of facts. Machines learn from such graphs by formulating a hypothesis and generating queries from it. Based on the results of the requests, we can evaluate how good the hypothesis is. Since we have very large knowledge graphs with several billion nodes and edges, it was important to process these inquiries as efficiently as possible.

We also deal with conversational AIs. A well-known example of this would be Siri or Alexa, but we want to be able to answer more complex questions. For this, natural language is processed and the relationships are mapped onto samples. These patterns can then be used to traverse knowledge graphs and find suitable answers.

Furthermore, we have been dealing with data integration for almost a decade. It is about bringing together large amounts of data through efficient and effective methods. We are interested in the question of how to bring complex databases together in such a way that answering increasingly complex questions becomes possible. Due to the enormous size of the data records, we first have to solve runtime problems. It is also a matter of ensuring that the links between the data are correct. Ultimately, these are machine learning problems. As such, we have developed a number new methods and algorithms to deal with this problem. We have also been able to demonstrate certain theoretical properties that have not previously been proven.

What does such an award mean for a scientist?

Of course you are happy because you invest body and soul in designing your research so that you have an effect, therefore impact. It's nice to see that you have had an impact, especially when this is confirmed by an external authority. It is also an incentive because you want to maintain, if not improve, your quality level.

Is there a milestone you absolutely want to reach?

I would like to see this idea of collaboration between machines and people implemented. Data-driven machines can tackle a lot of problems that we have not been able to tackle so far. But people still have this intuition, as Einstein called it, and machines don't. I would like to use this interplay of digital intelligence and human intelligence to solve relevant problems, e.g. can predict cancer-causing substances or potential medicines for diseases.

Has your work changed due to the Corona Virus Pandemic?   

I have the great advantage of being a computer scientist. So I just need a laptop and then I'm happy. With an internet connection, of course. In terms of research technology, something has changed in that you have more digital appointments. In terms of content, nothing has changed. In teaching, it is a change because you get less digital feedback from the students and still want to respond to the needs of the students. Lectures have to be prepared differently and the students have to be allowed to give feedback repeatedly.

The University for the Information Society