Gender Bias in Pretrained Swedish Embeddings

Magnus Sahlgren, Fredrik Olsson

[How to correct problems with metadata yourself]


Abstract
This paper investigates the presence of gender bias in pretrained Swedish embeddings. We focus on a scenario where names are matched with occupations, and we demonstrate how a number of standard pretrained embeddings handle this task. Our experiments show some significant differences between the pretrained embeddings, with word-based methods showing the most bias and contextualized language models showing the least. We also demonstrate that the previously proposed debiasing method does not affect the performance of the various embeddings in this scenario.
Anthology ID:
W19-6104
Volume:
Proceedings of the 22nd Nordic Conference on Computational Linguistics
Month:
September–October
Year:
2019
Address:
Turku, Finland
Editors:
Mareike Hartmann, Barbara Plank
Venue:
NoDaLiDa
SIG:
Publisher:
Linköping University Electronic Press
Note:
Pages:
35–43
Language:
URL:
https://aclanthology.org/W19-6104
DOI:
Bibkey:
Cite (ACL):
Magnus Sahlgren and Fredrik Olsson. 2019. Gender Bias in Pretrained Swedish Embeddings. In Proceedings of the 22nd Nordic Conference on Computational Linguistics, pages 35–43, Turku, Finland. Linköping University Electronic Press.
Cite (Informal):
Gender Bias in Pretrained Swedish Embeddings (Sahlgren & Olsson, NoDaLiDa 2019)
Copy Citation:
PDF:
https://preview.aclanthology.org/teach-a-man-to-fish/W19-6104.pdf