On December 9, Elena Zentgraf was honored with third prize at a ceremony held at the Rheinhessen Chamber of Industry and Commerce for her outstanding master’s thesis. Her thesis, titled “Indirect Metrics for Evaluating LLM-Based Information Extraction in Industry 4.0,” was written as part of the VAMoS Research College and was supervised by Prof. Dr. Jens Heidrich and Janek Groß.
Elena Zentgraf addressed a highly topical issue of great significance to industry: How can the quality of automatically generated submodels of the so-called “administrative shell” be reliably evaluated? The administrative shell forms the digital representation of a physical product and is a central concept of Industry 4.0. It enables production facilities, machines, or products to be described using standardized digital models and facilitates the exchange of information across companies.
With the increasing use of Large Language Models (LLMs), such as GPT-4 or DeepSeek, such models can be automatically generated from technical documentation. This holds enormous potential for the industry. At the same time, however, a new challenge arises: How can the quality of automatically generated content be evaluated reliably and reproducibly?
Elena Zentgraf developed an innovative evaluation method that is immediately applicable and helps ensure the successful implementation of LLM technologies in production-related contexts.
We extend our warmest congratulations to Elena on this award!
