Trustworthy Human Language Technologies

The Trustworthy Human Language Technologies (TrustHLT) group conducts bacis research in two areas of natural language processing (NLP), namely in privacy-preserving NLP and legal NLP.

The TrustHLT group’s research on privacy in NLP focuses on developing methods and models to protect sensitive information in text processing tasks. Our work explores privacy-preserving techniques such as differential privacy, text rewriting, and private data generation in various NLP areas. We address challenges like balancing privacy with downstream performance, scalability, or linguistic quality.

 

In the domain of legal NLP, TrustHLT investigates approaches to enhance the understanding and analysis of legal documents. Our research includes tasks such as legal argument mining, summarizing legal texts, and answering legal questions from laypeople. Our interdisciplinary project aims at providing valuable insights in fields like court decision analysis and civil law procedures.