| Date: | 2024, January 23 |
| Time: | 11:30 a. m. |
| Place: | G29-018 |
| Author: | Langer, Stefan |
| Title: | Creating a Knowledge Graph from Chemical Literature for the SmartProSys Project |
The SmartProSys research initiative aims to replace fossil raw materials in chemical production
with renewable carbon sources, thus contributing to a carbon-neutral society.
Knowledge Graphs can help with this task as they provide a structured and interconnected
representation of information, allowing researchers to efficiently explore the landscape
of a scientific domain, navigate and view complex relationships and learn about
novel approaches and ideas.
However, creating them is a complex and time-consuming endeavor and while new scientific findings
are constantly being published on platforms like ChemRxiv or in journals such as the
International Journal of Molecular Sciences, there is hardly enough capacity
to adapt ontologies or knowledge graphs accordingly.
In this talk, I propose an algorithmic approach that uses token classification to extract
chemical substances and chemical roles from research articles and a language model
to create a knowledge graph that links this information directly to specific locations
in the text corpus.