Rubrika: Akce
Monday, 16 December 2024: EdUKate: Multilingual Learning Materials for Ukrainian Children in Czech Schools
Akce 10. 12. 2024
Lucie PolákováMartin Popel (ÚFAL MFF UK) EdUKate is a TAČR-funded project carried out in partnership with Fraus, the largest textbook publisher in the Czech Republic. It aims to enhance the content and accessibility of digital education in the Czech educational system, with particular focus on Ukrainian speakers. We are adapting recent machine translation methods (Charles […]
Monday, 9 December 2024: HiČKoK: History of Czech in Corpus Continuum
Akce 26. 11. 2024
Daniel Zeman (ÚFAL MFF UK)Jiří Pergler (UJČ AV ČR) We will present the ongoing TAČR project focused on morphological annotation of texts from all historical stages of the Czech language. The goal of the project is to connect text corpora of different periods, so far built independently at different institutes, and to enrich them with […]
Monday, 2 December 2024: Behavioral studies of LLMs
Akce 26. 11. 2024
Jiří Milička (ÚČNK FF UK) Texts generated by large language models (LLMs) usually simulate an anthropomorphic persona, which stems from the fact that most of the training data is produced by and deals with humans, but also because people are willing to anthropomorphize practically anything, and the model accommodates this tendency. So, does it make […]
Seminář o jazykových datech a technologiích
Akce 21. 11. 2024
Seminář o jazykových datech a technologiích, pořádaný v rámci evropského projektu „Language Data Space“, přinese pohled na aktuální trendy v oblasti umělé inteligence a jazykových technologií.
Monday, 25 November 2024 – MSDRT: A formal framework for a communication-theoretic semantics and pragmatics of natural language
Akce 13. 11. 2024
Hans Kamp (Universität Stuttgart) MSDRT (‘Mental State Discourse Representation Theory’) is an extension of DRT that was originally designed to provide a semantic treatment of multiple attitude attributions, like ‘John believes there is a new drug against high blood pressure and he wants to get hold of it as soon as possible.’ This can be […]
Monday, 18 November 2024 – The Unnatural Language of Poetic Meters, or Why I Am Scared of SystematicityMonday
Akce 13. 11. 2024
Artjoms Šeļa (Institute of Czech Literature, Czech Academy of Sciences) While poetic meters are shaped and constrained by language, they also are external to it. No one speaks to their coworkers in iambic pentameter. Meters function as cultural technologies, imposing a systematic pattern on a language that is itself already full of structured relationships. The […]
Monday, 14 October, 2024 – Breaking the Curse of Multilinguality in Language Models
Akce 1. 10. 2024
While language models (LMs) grow larger and gain new capabilities, their performance in non-English languages increasingly lags behind. This is due to the curse of multilinguality, where each individual language’s performance suffers when models are trained on more languages. In this talk, I first examine how current language models do and don’t capture different languages […]
Monday, 7 October, 2024 – Cognitive insights into analysis of translationese combining product and process data
Akce 1. 10. 2024
The talk will be focused on „translationese,“ which refers to the specific linguistic properties of translation that distinguish it from non-translated language production. While these phenomena have been extensively analyzed in both human and machine translation studies, the driving forces behind translationese effects such as simplification and explicitation remain under-researched.
Linguistic Mondays and Fred Jelinek Seminar series
Akce 8. 4. 2024
We invite you to four lectures that will take place in April as part of Linguistic Mondays and Fred Jelinek Seminar series organized by the Institute of Formal and Applied Linguistics MFF UK. 8 April 2024Šárka Zikánová (ÚFAL MFF UK)Textová koherence ve starší češtiněhttps://ufal.mff.cuni.cz/node/2680 15 April 2024Kateřina Rysová (ÚFAL MFF UK)Čeština jako cizí jazyk: cesta […]
Multi-Source Simultaneous Speech Translation
Akce 26. 2. 2024
We investigate the opportunity to use multiple parallel speech signals — the original and simultaneous interpreting — as sources for translation to achieve higher quality of the simultaneous speech translation. We create an evaluation set ESIC (Europarl Simultaneous Interpreting Corpus). We analyze the challenges of simultaneous interpreting when used as an additional parallel source. Then, […]