Thomas Marwitz

Researchers led by Thomas Marwitz have demonstrated a method to predict emerging research directions in materials science by combining large language models with concept graphs built from scientific abstracts. The team trained a machine learning model on historical data to identify novel topic combinations that could inspire new research directions. The approach enables materials science experts to discover non-obvious research suggestions by analyzing semantic relationships in the literature. This work shows practical application of LLMs beyond text generation, using them to structure domain knowledge and forecast scientific trends.