Aleš Manuel Papáček
Main Research Interests
LLMs, RAG, Machine Translation, Etymology, Morphology
Projects
CZDEMOS4AI - A beneficial multi-agent AI avatar in a small democratic society.
- Project to create "AI avatars" (LLM with RAG) for various topics (history, law, media-security, addictology)
- I focus on evaluation of model outputs
Universal Segmentations - collection of lexical resources that captures morphological segmentations, harmonized into a cross-linguistically consistent annotation scheme.
Identifing Indo European roots in Czech words
Morph Origin Identification - etymology on level of individual morphs rather than whole words
LLM Judge Judge Club - Judge and validate your automated judges (see: "Do LLMs make more mistakes if they dont believe the input data?")
Curriculum Vitae
2025-... Mgr. in Arteficial Inteligence, MFF UK
2022-2025 Bc. in Computer Science (specialization: Artificial Intelligence), MFF UK
Teaching
NMAI059 - Probability and Statistics 1
- Summer semester 2026
- Information for students
Selected Bibliography
Publications:
- CUNI-UFAL at WMT26 Multilingual Instruction Shared Task. In: Proceedings of the Eleventh Conference on Machine Translation, pp. 1949-1958, Association for Computational Linguistics, Kerrville, TX, USA, ISBN 979-8-89176-474-3 (url, bibtex)
- CUNI-ÚFAL at the WMT26 Terminology Translation Task. In: Proceedings of the Eleventh Conference on Machine Translation, pp. 1849-1859, Association for Computational Linguistics, Kerrville, TX, USA, ISBN 979-8-89176-474-3 (url, bibtex)
- Tracing Morph Origins in Czech: A Computational Approach to Morph-Level Etymology. In: LREC 2026 | Fourth Workshop on Language Technologies for Historical and Ancient Languages (LT4HALA 2026) @ LREC 2026 Workshop Proceedings, pp. 7-17, anguage Resources Association (ELRA), Paris, France, ISBN 978-2-493814-58-6 (url, bibtex)
Datasets/software:
- Universal Segmentations 1.5 (DataSW). (url)
- DeriNet 2.3 (DataSW). (url)


