Chair of Computational Linguistics

Prof. Dr. Katja Markert

Contact Details

Institut für Computerlinguistik,
Universität Heidelberg
Im Neuenheimer Feld 325
69120 Heidelberg, Germany

Office hours	Thursday, 09.00–11.00
Phone	+49 6221 54-3245 (Secretary)
E-mail	markertcl.uni-heidelberg.de

Research Interest

I am working in the areas of trustworthy NLP, discourse, semantics, pragmatics and summarization. With regard to trustworthy NLP I care mostly about bias in NLP systems and language models as well as faithfulness. In the area of discourse I have long-standing interests in anaphora resolution beyond coreference, such as bridging and comparative anaphora. In semantics and pragmatics I concentrate on sentiment analysis and figurative language. Recent work in summarization focuses on the summarization of large document collections on long-running events, so-called timeline summarization as well as summarization evaluation.

I am always looking for new students to work on the above topics. Potential thesis topics for Bachelor and MA theses can be found under Thesis Topics.

Group Members, Recent and Current Collaborators and Alumni

• Julius Steen, PhD student
• Esther Van den Berg, Phd Student
• Prof. Michael Wiegand
• Ines Reinig
• Dr. Yufang Hou

I also work regularly with colleagues at the Natural Language Group at the Heidelberg Institute of Theoretical Studies as well as the L3S Web Science Research Centre at the University of Hanover.

Until 2021, I was co-director of the Leibniz-Science-Campus "Empirical Linguistics and Computational Language Modeling" where we concentrated on NLP models for the German language, in a joint project between Heidelberg University and the Institute for the German Language. I was PI on Area A, together with Prof. Dr Andreas Witt. The area focused on bias and framing.

News and Openings

• Looking for a new PhD student in the area of trustworthy NLP starting October 2024.
• New Paper on Trustworthy NLP on ArXiv.
  --- Steen, J; Markert, K. (2024): Gender Bias in News Summarization: Measures, Pitfalls and Corpora.
  
• New Publication on Trustworthy NLP at ACL 2023!
  -- Steen, J; Opitz, J; Frank, A; Katja Markert: With a Little Push, NLI Models can Robustly and Efficiently predict Faithfulness
• New Publication on Language Model Evaluation at IWCS 2023!
  --- Reinig, I; Markert, K: Can current NLI systems handle German word order? Investigating language model performance on a new German challenge set of minimal pairs.
• New Publication on Summarization!
  --- Fatima, M,; Kolber, T; Markert, K; Strube, M. (2023) SimCSum: Joint Learning of Simplification and Cross-Lingual Summarization for Cross-Lingual Science Journalism