Workshop: Structured Information Extraction with LLMs

🗓️ Event	General Online Research 2025
📅 Date	31 March 2025
📍 Location	Freie Universität Berlin
🌐 Language	English
📥 Materials	Slides, Jupyter Notebook

This 2.5-hour hands-on workshop introduced participants to structured information extraction using Large Language Models (LLMs). Through practical exercises and discussions, participants learned:

1. How to extract structured information (categories, entities, relations) from text and images using LLMs
2. The tradeoffs between traditional NLP approaches and LLM-based extraction
3. Techniques to improve extraction accuracy through prompt engineering and few shot learning
4. The landscape of models and libraries for structured LLM responses
5. Best practices for monitoring and deploying LLM-based systems in production

The workshop included a micro-hackathon where participants competed to achieve the highest accuracy on the AGNews classification task.

The workshop built on the GOR24 presentation Where do LLMs fit in NLP pipelines? by my colleague Paavo Huvoiala and me.