Hi! I’m Xaver, a first year PhD student and research assistant at the Chair of Computational Linguistics at the University of Augsburg, supervised by Prof. Annemarie Friedrich. My research is broadly based on natural language processing in low-resource settings, building up a particular interest in understanding human vs. AI generated narration and improving NLP systems and generative AI-powered applications in this regard.

Before starting my PhD and directly after my Masters, I worked as a Junior Data Scientist at CHECK24 for one and a half years, where I improved and participated in the development of RAG-based chatbot and mailbot systems, agentic LLM frameworks for automated customer service, and generative AI components into production dialogue pipelines.

Research Interests

My academic interests sit at the intersection of low-resource NLP, language generation systems, and narrativity. I am broadly working on:

• Low-resource NLP — robust models for languages and varieties with limited training data
• Narrative detection, fact checking & misinformation detection — a new approach to linguistic operationalisation of narrativity as a possible detection signal for misleading content; understanding how information is framed and verified in text
• Human-centric and trustworthy NLP — building systems that are reliable, interpretable, and fair
• NLP for dialects and non-standard languages — extending the reach of NLP tools beyond standard written language (a fable from my Master’s thesis)

A question that ties these threads together could be:What makes a text a narrative — and can we detect the linguistic fingerprints of narration well enough to surface misleading stories in texts where we would not expect to find them?

Background

I hold an M.Sc. in Computational Linguistics (grade 1.11) and a B.Sc. in Computational Linguistics (grade 1.58), both from Ludwig-Maximilians-University Munich, as well as a B.A. in English Studies. My master’s thesis in computational linguistics investigated multiple zero-shot transfer learning methods for slot and intent detection in Upper German dialects whilst my bachelor’s thesis focused on creating a multilingual gold standard for case marker extraction.