Probing Pre-Trained Language Models for Cross-Cultural Differences in Values
Research output: Working paper › Preprint › Research
Standard
Probing Pre-Trained Language Models for Cross-Cultural Differences in Values. / Arora, Arnav; Kaffee, Lucie-Aimée Frimelle; Augenstein, Isabelle.
arxiv.org, 2022.Research output: Working paper › Preprint › Research
Harvard
APA
Vancouver
Author
Bibtex
}
RIS
TY - UNPB
T1 - Probing Pre-Trained Language Models for Cross-Cultural Differences in Values
AU - Arora, Arnav
AU - Kaffee, Lucie-Aimée Frimelle
AU - Augenstein, Isabelle
PY - 2022
Y1 - 2022
N2 - Language embeds information about social, cultural, and political values people hold. Prior work has explored social and potentially harmful biases encoded in Pre-Trained Language models (PTLMs). However, there has been no systematic study investigating how values embedded in these models vary across cultures. In this paper, we introduce probes to study which values across cultures are embedded in these models, and whether they align with existing theories and cross-cultural value surveys. We find that PTLMs capture differences in values across cultures, but those only weakly align with established value surveys. We discuss implications of using mis-aligned models in cross-cultural settings, as well as ways of aligning PTLMs with value surveys.
AB - Language embeds information about social, cultural, and political values people hold. Prior work has explored social and potentially harmful biases encoded in Pre-Trained Language models (PTLMs). However, there has been no systematic study investigating how values embedded in these models vary across cultures. In this paper, we introduce probes to study which values across cultures are embedded in these models, and whether they align with existing theories and cross-cultural value surveys. We find that PTLMs capture differences in values across cultures, but those only weakly align with established value surveys. We discuss implications of using mis-aligned models in cross-cultural settings, as well as ways of aligning PTLMs with value surveys.
U2 - 10.48550/arXiv.2203.13722
DO - 10.48550/arXiv.2203.13722
M3 - Preprint
BT - Probing Pre-Trained Language Models for Cross-Cultural Differences in Values
PB - arxiv.org
ER -
ID: 339345259