Master’s Thesis - LLM-Based Moving-Target Detection and Signal Generation in Economic Narratives

Stellenbeschreibung:

Location: ERL S SK 3

Mode of Employment: Full-time / Fixed Term

A Snapshot of Your Day

You work with public economic and policy-related text and investigate how recurring communication changes over time. exploring LLM-based methods for identifying newly emphasized, de-emphasized, or disappearing themes and exploring how such changes can be transformed into structured and interpretable signal.

How You’ll Make An Impact

Development of a research approach for temporal comparison of recurring economic or policy communication
Exploration of LLM-based methods for identifying newly emphasized, de-emphasized, or disappearing themes
Implementation of a prototype for structured signal generation from changes in narrative emphasis
Evaluation of alternative methodological approaches for temporal text comparison and signal stability
Documentation of methodology, findings, and limitations for future reuse in applied text analytics
Support with the translation of research results into interpretable signal concepts for forecasting‑related use cases

What You Bring

Studies in Data Science, Computer Science, Artificial Intelligence, or a related field at Master’s level
Knowledge of Python-based data processing and NLP workflows for text analysis and prototyping
Interest in large language models, prompt‑based methods, and temporal text analysis in economic or business‑related contexts
Understanding of research‑oriented experimentation and evaluation design for comparing methodological alternatives
Ability to work independently on implementation, analysis, and technical documentation in an applied research environment
Experience with embeddings, prompt‑based extraction, topic modeling, or text clustering as an advantage