Alexander Loeser

Alexander Loeser is a researcher active in natural language processing (NLP) and its applications in clinical and financial domains. He has contributed to diverse areas including clinical outcome prediction, transformer-based reinforcement learning environments, domain knowledge integration, and information extraction from text. His recent work focuses on evaluating large language models' financial literacy via domain-specific languages and addressing data drift in clinical NLP tasks.

• Key Research Areas:
  • Clinical decision support systems and outcome prediction
  • Domain knowledge injection into transformer models
  • Biased news article detection
  • Interactive NLP systems for entity linking
• Methodological Focus:
  • Reinforcement learning and attention mechanisms
  • Multi-task and self-supervised learning
  • Active sampling for annotation efficiency
  • Topic segmentation and classification

Loeser has collaborated extensively with researchers like Wolfgang Nejdl, Betty van Aken, Felix Gers, and Paul Grundmann, with publications spanning from 2012 to 2025. His work emphasizes interpretability, generalization, and practical deployment of NLP models in real-world domains.