Prof. Jennifer Strunk reports
International Collaboration: Prof. Jennifer Strunk’s Insights from the Collaboration with Hokkaido University
We are delighted to share the invaluable experiences of Prof. Jennifer Strunk during her collaboration with Prof. Keisuke Takahashi in Japan. Through her partnership with the esteemed Hokkaido University, Prof. Strunk has made a significant contribution to advancing the goals of NFDI4Cat on an international scale. Join us as we explore the enriching outcomes of this fruitful cooperation!
In December 2022, Prof. Jennifer Strunk and Dr. Abdo Hezam Mohsen traveled to Sapporo for one week to interlink the activities in NFDI4Cat with Prof. Keisuke Takahashi, who is one of the leading international experts in the field of “Data Science in Catalysis”.
The aim of the visit was to compare the ontologies and databases developed and to be further developed in NFDI4Cat with the ongoing databases of the Takahashi group, using the example of “halide perovskites” in photocatalytic applications. However, the latter are not thematically broad, as is the goal in NFDI4Cat, but relate very closely to the research topic of heterogeneously catalysed methane activation. The approach in NFDI4Cat can consequently be described as “top down”, whereas here a more “bottom up” approach was followed.
In Japan, the collaboration followed a specific division of labor: Jennifer Strunk worked on ontology development with Dr. Lauren Takahashi, focusing on their expertise. Meanwhile, Abdo Mohsen teamed up with Keisuke Takahashi to handle database work, including data extraction and integration into machine learning. The topic of photocatalysis brought by the German team was novel and introduced fresh perspectives to our Japanese partners.
During the collaboration, a profound understanding of each other’s perspectives was achieved, and the ontology development was effectively visualized. Abdo Mohsen learned the structure of the database, the automated extraction of the data, and he applied it to an Excel spreadsheet he brought with him based on literature data on the photocatalytic activity of halide perovskites. The machine learning itself failed, but this led to the crucial realization that the data reported nowadays in scientific publications are insufficient. There is a lack of negative results, and insufficient detail in the reporting of experimental parameters. Both are important findings to improve the new databases in NFDI4Cat. A publication of the results in a short article (e.g., as a perspective article) is planned.
“The machine learning itself failed, but this led to the crucial realization that the data reported nowadays in scientific publications are insufficient. There is a lack of negative results, and insufficient detail in the reporting of experimental parameters. Both are important findings to improve the new databases in NFDI4Cat.”
– Frau Prof. Jennifer Strunk
Opportunities for specific international exchange are an asset for NFDI4Cat, and the currently advanced development of activities in this area is very positively evaluated. The stay in Japan would not have been possible without the support of NFDI4Cat. Although we developed the suspicion that machine evaluation of literature databases might fail because of the quality of the data itself already before our stay, we were now able to verify it in this collaboration. This laid the foundation to do better in NFDI4Cat. The mutual understanding of scientists in the fields of thermal catalysis (Japan) and photocatalysis (Germany) also helped because the Japanese databases did not include photon-driven processes. The broadening of the knowledge base thus achieved on both sides also advances the field of “data science” as such.