Varvara Arzt

Varvara Arzt is a PreDoc Researcher in the Department of Data Science at TU Wien's Faculty of Informatics. She specializes in Natural Language Processing (NLP), Information Retrieval, and Computational Linguistics. She teaches courses such as Natural Language Processing and Information Extraction, Advanced Information Retrieval, and Innovation. Her research focuses on improving transparency in NLP benchmarks, analyzing biographical data through language models, and exploring digital humanities applications like automated analysis of music lyrics. She leads the project 'Designing an Annotation Framework for Creating a Health Dataset from Layperson Medical Forums in Bahasa Indonesia' funded by the Austrian Exchange Service (OeAD).

Her recent publications address critical issues like hallucination detection in text generation models and the generalization limits of language models in relation extraction tasks. She advocates for methodological rigor in benchmark creation and leaderboards. Varvara has no listed scientific awards but actively contributes to academic discourse through peer-reviewed articles and teaching. She coordinates the mentioned projects and collaborates on interdisciplinary research spanning computational linguistics and cultural studies.