Leveraging “Commonalities” in low-resource setting

Organisatsiooni nimi
TartuNLP
Kokkuvõte
In this thesis, the goal is to transfer the NLP capabilities from high-resourced languages to a low-resourced language settings. In particular, this work leverages the transformer-based pretrained models and investigates the knowledge transfer possibilities on related languages, distant languages, and within the target (low-resourced) language.
The focus is on Estonian (as the target language) and on the Sentiment Analysis task.
Lõputöö kaitsmise aasta
2023-2024
Juhendaja
Somnath Banerjee, Siim Orasmaa
Suhtlemiskeel(ed)
eesti keel, inglise keel
Nõuded kandideerijale
Tase
Bakalaureus, Magister
Märksõnad
#NLP, #LanguageModels, #Sentiment Analysis

Kandideerimise kontakt

 
Nimi
Somanth Banerjee, Siim Orasmaa
Tel
E-mail
somnath.banerjee@ut.ee; siim.orasmaa@ut.ee
Vaata lähemalt
http://www.ut.ee//~somnath/