Unsupervised Mitigation of Gender Bias by Character Components: A Case Study of Chinese Word Embedding

Xiuying Chen; Mingzhe Li; Rui Yan; Xin Gao; Xiangliang Zhang

Unsupervised Mitigation of Gender Bias by Character Components: A Case Study of Chinese Word Embedding

Xiuying Chen, Mingzhe Li, Rui Yan, Xin Gao, Xiangliang Zhang^*

^*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

4 Scopus citations

Abstract

Word embeddings learned from massive text collections have demonstrated significant levels of discriminative biases. However, debiasing on the Chinese language, one of the most spoken languages, has been less explored. Meanwhile, existing literature relies on manually created supplementary data, which is time- and energy-consuming. In this work, we propose the first Chinese Gender-neutral word Embedding model (CGE) based on Word2vec, which learns gender-neutral word embeddings without any labeled data. Concretely, CGE utilizes and emphasizes the rich feminine and masculine information contained in radicals, i.e., a kind of component in Chinese characters, during the training procedure. This consequently alleviates discriminative gender biases. Experimental results show that our unsupervised method outperforms the state-of-the-art supervised debiased word embedding models without sacrificing the functionality of the embedding model.

Original language	English (US)
Title of host publication	GeBNLP 2022 - 4th Workshop on Gender Bias in Natural Language Processing, Proceedings of the Workshop
Editors	Christian Hardmeier, Christian Hardmeier, Christine Basta, Basta Christine, Marta R. Costa-Jussa, Gabriel Stanovsky, Hila Gonen
Publisher	Association for Computational Linguistics (ACL)
Pages	121-128
Number of pages	8
ISBN (Electronic)	9781955917681
State	Published - 2022
Event	4th Workshop on Gender Bias in Natural Language Processing, GeBNLP 2022 - Seattle, United States Duration: Jul 15 2022 → …

Publication series

Name	GeBNLP 2022 - 4th Workshop on Gender Bias in Natural Language Processing, Proceedings of the Workshop

Conference

Conference	4th Workshop on Gender Bias in Natural Language Processing, GeBNLP 2022
Country/Territory	United States
City	Seattle
Period	07/15/22 → …

Bibliographical note

Publisher Copyright:
© 2022 Association for Computational Linguistics.

ASJC Scopus subject areas

Computational Theory and Mathematics
Computer Science Applications
Information Systems
General Psychology
Gender Studies

Cite this

Chen, X., Li, M., Yan, R., Gao, X., & Zhang, X. (2022). Unsupervised Mitigation of Gender Bias by Character Components: A Case Study of Chinese Word Embedding. In C. Hardmeier, C. Hardmeier, C. Basta, B. Christine, M. R. Costa-Jussa, G. Stanovsky, & H. Gonen (Eds.), GeBNLP 2022 - 4th Workshop on Gender Bias in Natural Language Processing, Proceedings of the Workshop (pp. 121-128). (GeBNLP 2022 - 4th Workshop on Gender Bias in Natural Language Processing, Proceedings of the Workshop). Association for Computational Linguistics (ACL).

Chen, Xiuying ; Li, Mingzhe ; Yan, Rui et al. / Unsupervised Mitigation of Gender Bias by Character Components : A Case Study of Chinese Word Embedding. GeBNLP 2022 - 4th Workshop on Gender Bias in Natural Language Processing, Proceedings of the Workshop. editor / Christian Hardmeier ; Christian Hardmeier ; Christine Basta ; Basta Christine ; Marta R. Costa-Jussa ; Gabriel Stanovsky ; Hila Gonen. Association for Computational Linguistics (ACL), 2022. pp. 121-128 (GeBNLP 2022 - 4th Workshop on Gender Bias in Natural Language Processing, Proceedings of the Workshop).

@inproceedings{38ff3dffb6b74a4c9c9c8bd8b35dd740,

title = "Unsupervised Mitigation of Gender Bias by Character Components: A Case Study of Chinese Word Embedding",

abstract = "Word embeddings learned from massive text collections have demonstrated significant levels of discriminative biases. However, debiasing on the Chinese language, one of the most spoken languages, has been less explored. Meanwhile, existing literature relies on manually created supplementary data, which is time- and energy-consuming. In this work, we propose the first Chinese Gender-neutral word Embedding model (CGE) based on Word2vec, which learns gender-neutral word embeddings without any labeled data. Concretely, CGE utilizes and emphasizes the rich feminine and masculine information contained in radicals, i.e., a kind of component in Chinese characters, during the training procedure. This consequently alleviates discriminative gender biases. Experimental results show that our unsupervised method outperforms the state-of-the-art supervised debiased word embedding models without sacrificing the functionality of the embedding model.",

author = "Xiuying Chen and Mingzhe Li and Rui Yan and Xin Gao and Xiangliang Zhang",

note = "Publisher Copyright: {\textcopyright} 2022 Association for Computational Linguistics.; 4th Workshop on Gender Bias in Natural Language Processing, GeBNLP 2022 ; Conference date: 15-07-2022",

year = "2022",

language = "English (US)",

series = "GeBNLP 2022 - 4th Workshop on Gender Bias in Natural Language Processing, Proceedings of the Workshop",

publisher = "Association for Computational Linguistics (ACL)",

pages = "121--128",

editor = "Christian Hardmeier and Christian Hardmeier and Christine Basta and Basta Christine and Costa-Jussa, {Marta R.} and Gabriel Stanovsky and Hila Gonen",

booktitle = "GeBNLP 2022 - 4th Workshop on Gender Bias in Natural Language Processing, Proceedings of the Workshop",

address = "United States",

}

Chen, X, Li, M, Yan, R, Gao, X & Zhang, X 2022, Unsupervised Mitigation of Gender Bias by Character Components: A Case Study of Chinese Word Embedding. in C Hardmeier, C Hardmeier, C Basta, B Christine, MR Costa-Jussa, G Stanovsky & H Gonen (eds), GeBNLP 2022 - 4th Workshop on Gender Bias in Natural Language Processing, Proceedings of the Workshop. GeBNLP 2022 - 4th Workshop on Gender Bias in Natural Language Processing, Proceedings of the Workshop, Association for Computational Linguistics (ACL), pp. 121-128, 4th Workshop on Gender Bias in Natural Language Processing, GeBNLP 2022, Seattle, United States, 07/15/22.

Unsupervised Mitigation of Gender Bias by Character Components: A Case Study of Chinese Word Embedding. / Chen, Xiuying; Li, Mingzhe; Yan, Rui et al.
GeBNLP 2022 - 4th Workshop on Gender Bias in Natural Language Processing, Proceedings of the Workshop. ed. / Christian Hardmeier; Christian Hardmeier; Christine Basta; Basta Christine; Marta R. Costa-Jussa; Gabriel Stanovsky; Hila Gonen. Association for Computational Linguistics (ACL), 2022. p. 121-128 (GeBNLP 2022 - 4th Workshop on Gender Bias in Natural Language Processing, Proceedings of the Workshop).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Unsupervised Mitigation of Gender Bias by Character Components

T2 - 4th Workshop on Gender Bias in Natural Language Processing, GeBNLP 2022

AU - Chen, Xiuying

AU - Li, Mingzhe

AU - Yan, Rui

AU - Gao, Xin

AU - Zhang, Xiangliang

PY - 2022

Y1 - 2022

N2 - Word embeddings learned from massive text collections have demonstrated significant levels of discriminative biases. However, debiasing on the Chinese language, one of the most spoken languages, has been less explored. Meanwhile, existing literature relies on manually created supplementary data, which is time- and energy-consuming. In this work, we propose the first Chinese Gender-neutral word Embedding model (CGE) based on Word2vec, which learns gender-neutral word embeddings without any labeled data. Concretely, CGE utilizes and emphasizes the rich feminine and masculine information contained in radicals, i.e., a kind of component in Chinese characters, during the training procedure. This consequently alleviates discriminative gender biases. Experimental results show that our unsupervised method outperforms the state-of-the-art supervised debiased word embedding models without sacrificing the functionality of the embedding model.

AB - Word embeddings learned from massive text collections have demonstrated significant levels of discriminative biases. However, debiasing on the Chinese language, one of the most spoken languages, has been less explored. Meanwhile, existing literature relies on manually created supplementary data, which is time- and energy-consuming. In this work, we propose the first Chinese Gender-neutral word Embedding model (CGE) based on Word2vec, which learns gender-neutral word embeddings without any labeled data. Concretely, CGE utilizes and emphasizes the rich feminine and masculine information contained in radicals, i.e., a kind of component in Chinese characters, during the training procedure. This consequently alleviates discriminative gender biases. Experimental results show that our unsupervised method outperforms the state-of-the-art supervised debiased word embedding models without sacrificing the functionality of the embedding model.

UR - http://www.scopus.com/inward/record.url?scp=85137567959&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:85137567959

T3 - GeBNLP 2022 - 4th Workshop on Gender Bias in Natural Language Processing, Proceedings of the Workshop

SP - 121

EP - 128

BT - GeBNLP 2022 - 4th Workshop on Gender Bias in Natural Language Processing, Proceedings of the Workshop

A2 - Hardmeier, Christian

A2 - Basta, Christine

A2 - Christine, Basta

A2 - Costa-Jussa, Marta R.

A2 - Stanovsky, Gabriel

A2 - Gonen, Hila

PB - Association for Computational Linguistics (ACL)

Y2 - 15 July 2022

ER -

Chen X, Li M, Yan R, Gao X, Zhang X. Unsupervised Mitigation of Gender Bias by Character Components: A Case Study of Chinese Word Embedding. In Hardmeier C, Hardmeier C, Basta C, Christine B, Costa-Jussa MR, Stanovsky G, Gonen H, editors, GeBNLP 2022 - 4th Workshop on Gender Bias in Natural Language Processing, Proceedings of the Workshop. Association for Computational Linguistics (ACL). 2022. p. 121-128. (GeBNLP 2022 - 4th Workshop on Gender Bias in Natural Language Processing, Proceedings of the Workshop).

Unsupervised Mitigation of Gender Bias by Character Components: A Case Study of Chinese Word Embedding

Abstract

Publication series

Conference

Bibliographical note

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this