PhD
Chair of Statistics and Data Science in Social Sciences and the Humanities at LMU
Social Data Science and AI Lab (SODA)
Department of Statistics
Ludwigstr. 33
80539 München
Biosketch
Developing and analysing NLP methods for climate domain. Background in computational linguistics, economics and statistics. Experience as data analyst, software developer and NLP engineer.
relAI Research
Her research applies large language models to extract carbon emissions data from unstructured corporate sustainability reports, as part of a collaboration between the German Federal Bank, LMU, and TUM. She has designed and evaluated a RAG-based extraction pipeline, examining how prompt design, context, and model choice affect performance. This work has produced a benchmark dataset published in Nature Scientific Data, a prompt evaluation study under review, and a contribution to an EACL 2026 survey on NLP for Social Good.
She is now extending this work in two directions. First, Anna is investigating uncertainty quantification methods—based on output consistency and inference-time reasoning traces—to provide users with meaningful reliability signals. Second, she is developing an agentic system to verify environmental claims against heterogeneous external sources such as emissions databases and satellite data, enabling cross-source validation of corporate disclosures in this high-stakes domain.