Browsing by Author "Bjerva, Johannes"

Now showing 1 - 4 of 4

listelement.badge.access-status Open Access ,
Colex2Lang: Language Embeddings from Semantic Typology
(University of Tartu Library, 2023-05) Chen, Yiyi; Biswas, Russa; Bjerva, Johannes
listelement.badge.access-status Open Access ,
Cross-lingual Learning of Semantic Textual Similarity with Multilingual Word Representations
(Gothenburg, Sweden, Association for Computational Linguistics, pp. 211--215, 2017) Bjerva, Johannes; Östling, Robert; Tiedemann, Jörg; Tahmasebi, Nina
listelement.badge.access-status Open Access ,
Tokenization on Trial: The Case of Kalaallisut–Danish Legal Machine Translation
(University of Tartu Library, 2025-03) Ploeger, Esther; Saucedo, Paola; Bjerva, Johannes; Kristensen-McLachlan, Ross Deans; Lent, Heather; Johansson, Richard; Stymne, Sara
The strengths of subword tokenization have been widely demonstrated when applied to higher-resourced, morphologically simple languages. However, it is not self-evident that these results transfer to lower-resourced, morphologically complex languages. In this work, we investigate the influence of different subword segmentation techniques on machine translation between Danish and Kalaallisut, the official language of Greenland. We present the first semi-manually aligned parallel corpus for this language pair, and use it to compare subwords from unsupervised tokenizers and morphological segmenters. We find that Unigram-based segmentation both preserves morphological boundaries and handles out-of-vocabulary words adequately, but that this does not directly correspond to superior translation quality. We hope that our findings lay further groundwork for future efforts in neural machine translation for Kalaallisut.
listelement.badge.access-status Open Access ,
Will my auxiliary tagging task help? Estimating Auxiliary Tasks Effectivity in Multi-Task Learning
(Gothenburg, Sweden, Association for Computational Linguistics, pp. 216--220, 2017) Bjerva, Johannes; Tiedemann, Jörg; Tahmasebi, Nina