معرفی
David Mareček is a researcher at the Institute of Formal and Applied Linguistics (ÚFAL) within the Faculty of Mathematics and Physics at Charles University. His work focuses on neural network interpretation, machine learning, dependency syntax analysis, and machine translation, with a particular emphasis on unsupervised and semi-supervised parsing techniques.
- Current projects: GenderBias (2023–present), EduPo (2024–present)
- Long-term contributions in dependency treebanks (HamleDT, 2012–2016) and neural machine translation frameworks (Treex, TectoMT)
His research explores linguistic structure in neural architectures, including BERT and Transformer models, through methods like structural probing and attention analysis. Recent publications examine gender bias mitigation, automated poetry analysis, and theatre script generation.
Key collaborations include Rudolf Rosa, Tomáš Musil, and Tomasz Limisiewicz. He supervises students in projects spanning multilingual representations, language model probing, and bias detection in AI systems.
Technical leadership in developing tools like THEaiTRobot and DEPFIX demonstrates his commitment to applied computational linguistics and robust NLP systems.
