Quality and Usability

XplaiNLP

Goal

Create intelligent decision support systems, by researching the whole cycle from developing and implementing large language models, and designing user interfaces with human-meaningful representations of model outputs and metadata, by implementing explanations and transparency features from NLP-based predictions.

Approach: 

  • Develop and apply LLMs for fake news detection in text data 
  • Develop and apply LLMs for Sustainable Development Goal detection in different text sources 
  • Design and validate intelligent decision-support systems for fake news detection 
  • Implement and validate human-meaningful explanations for the fake news detection task  

Teaching: 

  • NLP Seminar (summer term) 
  • Ethics of AI Seminar (summer term)
  • Advanced Study Projects (summer and winter term)

Projects: 

Topics

  • Fake news detection (text and image) 
  • Hate speech detection 
  • Claim extraction and claim verification 
  • NLP downstream tasks 
  • Sustainable Development Goal detection in text 
  • Human-Computer Interaction 
  • Explainable AI 
  • AI regulation (impact analysis of AI Act on LLMs) 
  • Bias in dataset and models

Team:

Senior Researcher

Sebastian Möller

Researchers: 

Partner

Students:

  • Ata Nizamoglu
  • Martin Burghart