Christopher Brückner

Main Research Interests

I am currently exploring graph neural networks for text classification with hierarchically ordered label taxonomies, as well as their applications in named entity disambiguation with hierarchical entities and other graph-constrained NLP problems. I focus on multilingual domain-specific solutions and texts that make people sad. Digital humanities, digital cultural heritage, some historical language change and biomedical NLP.

If you want to learn more about how graphs are related to modern NLP, I recommend the following article: Transformers are Graph Neural Networks

For an introduction to GNNs, check out these Stanford slides: A General Perspective on Graph Neural Networks
 

Projects

Horizon Europe

• MEMORISE: Virtualisation and Multimodal Exploration of Heritage on Nazi Persecution (2024-2026)
• TWIN4DEM: Strengthening Democratic Resilience Through Digital Twins (2025-2026)

MŠMT

• GI-Insight: Improving stomach examinations with Artificial Intelligence: A deep learning approach for assisted gastroscopy (2024-2026)

Curriculum Vitae

Academia

• since 2023: Ph.D. in Computational Linguistics at ÚFAL MFF UK
  Topic: Information extraction from domain-specific data
  Supervisor: Pavel Pecina
• 2020 - 2023: M.Sc. in Data and Computer Science at Heidelberg University, Germany
  Thesis: Multi-Feature Clustering of Search Results (pdf, slides)
  Minor: Computational Linguistics
• 2016 - 2020: B.Sc. in Applied Computer Science at Heidelberg University, Germany
  Thesis: Structure-Centric Near-Duplicate Detection (pdf, slides)
  Minor: Physics

Industry

• 2022 - 2023: Data Scientist at sovanta AG, Heidelberg, Germany
  Multilingual text classification and anomaly detection for imbalanced data in cloud environments

Teaching

• 2022 - 2023: Mining Massive Datasets at Heidelberg University, Germany
  Based on the Stanford University course CS246 (slides)

Selected Bibliography

• Google Scholar
• ORCID: 0009-0003-1998-4097
• Scopus ID: 58948296200

1. Christopher Brückner and Pavel Pecina. 2025. Towards Semantic Tagging of Segmented Holocaust Narratives. In Compendium of Papers of the Prague Visual History and Digital Humanities Conference 2025, pages 177–192, Prague, Czechia. MatfyzPress. (pdf, slides)
2. Christopher Brückner, Leixin Zhang, and Pavel Pecina. 2024. Similarity-Based Cluster Merging for Semantic Change Modeling. In Proceedings of the 5th Workshop on Computational Approaches to Historical Language Change, pages 23–28, Bangkok, Thailand. Association for Computational Linguistics. (poster)