@inproceedings{chen-etal-2023-structural,
title = "Structural Contrastive Pretraining for Cross-Lingual Comprehension",
author = "Chen, Nuo and
Shou, Linjun and
Song, Tengtao and
Gong, Ming and
Pei, Jian and
Chang, Jianhui and
Jiang, Daxin and
Li, Jia",
editor = "Rogers, Anna and
Boyd-Graber, Jordan and
Okazaki, Naoaki",
booktitle = "Findings of the Association for Computational Linguistics: ACL 2023",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-acl.128",
doi = "10.18653/v1/2023.findings-acl.128",
pages = "2042--2057",
abstract = "To present, multilingual language models trained using various pre-training tasks like mask language modeling (MLM) have yielded encouraging results on a wide range of downstream tasks. Despite the promising performances, structural knowledge in cross-lingual corpus is less explored in current works, leading to the semantic misalignment. In this paper, we propose a new pre-training task named Structural Contrast Pretraining (SCP) to align the structural words in a parallel sentence, enhancing the models{'} ability to comprehend cross-lingual representations. Concretely, each structural word in source and target languages is regarded as a positive pair in SCP. Since contrastive learning compares positive and negative pairs, an increase in the frequency of negative pairings could enhance the performance of the resulting model. Therefore, we further propose Cross-lingual Momentum Contrast (CL-MoCo) to increase the number of negative pairs by maintaining a large size of the queue. CL-MoCo extends the original Moco approach into cross-lingual training and jointly optimizes the source-to-target language and target-to-source language representations, resulting in a more suitable encoder for cross-lingual transfer. We conduct extensive experiments to validate the proposed approach on three cross-lingual tasks across five datasets such as MLQA, WikiAnn, etc, and results prove the effectiveness of our method.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="chen-etal-2023-structural">
<titleInfo>
<title>Structural Contrastive Pretraining for Cross-Lingual Comprehension</title>
</titleInfo>
<name type="personal">
<namePart type="given">Nuo</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Linjun</namePart>
<namePart type="family">Shou</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tengtao</namePart>
<namePart type="family">Song</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ming</namePart>
<namePart type="family">Gong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jian</namePart>
<namePart type="family">Pei</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jianhui</namePart>
<namePart type="family">Chang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Daxin</namePart>
<namePart type="family">Jiang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jia</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2023-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: ACL 2023</title>
</titleInfo>
<name type="personal">
<namePart type="given">Anna</namePart>
<namePart type="family">Rogers</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jordan</namePart>
<namePart type="family">Boyd-Graber</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Naoaki</namePart>
<namePart type="family">Okazaki</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Toronto, Canada</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>To present, multilingual language models trained using various pre-training tasks like mask language modeling (MLM) have yielded encouraging results on a wide range of downstream tasks. Despite the promising performances, structural knowledge in cross-lingual corpus is less explored in current works, leading to the semantic misalignment. In this paper, we propose a new pre-training task named Structural Contrast Pretraining (SCP) to align the structural words in a parallel sentence, enhancing the models’ ability to comprehend cross-lingual representations. Concretely, each structural word in source and target languages is regarded as a positive pair in SCP. Since contrastive learning compares positive and negative pairs, an increase in the frequency of negative pairings could enhance the performance of the resulting model. Therefore, we further propose Cross-lingual Momentum Contrast (CL-MoCo) to increase the number of negative pairs by maintaining a large size of the queue. CL-MoCo extends the original Moco approach into cross-lingual training and jointly optimizes the source-to-target language and target-to-source language representations, resulting in a more suitable encoder for cross-lingual transfer. We conduct extensive experiments to validate the proposed approach on three cross-lingual tasks across five datasets such as MLQA, WikiAnn, etc, and results prove the effectiveness of our method.</abstract>
<identifier type="citekey">chen-etal-2023-structural</identifier>
<identifier type="doi">10.18653/v1/2023.findings-acl.128</identifier>
<location>
<url>https://aclanthology.org/2023.findings-acl.128</url>
</location>
<part>
<date>2023-07</date>
<extent unit="page">
<start>2042</start>
<end>2057</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Structural Contrastive Pretraining for Cross-Lingual Comprehension
%A Chen, Nuo
%A Shou, Linjun
%A Song, Tengtao
%A Gong, Ming
%A Pei, Jian
%A Chang, Jianhui
%A Jiang, Daxin
%A Li, Jia
%Y Rogers, Anna
%Y Boyd-Graber, Jordan
%Y Okazaki, Naoaki
%S Findings of the Association for Computational Linguistics: ACL 2023
%D 2023
%8 July
%I Association for Computational Linguistics
%C Toronto, Canada
%F chen-etal-2023-structural
%X To present, multilingual language models trained using various pre-training tasks like mask language modeling (MLM) have yielded encouraging results on a wide range of downstream tasks. Despite the promising performances, structural knowledge in cross-lingual corpus is less explored in current works, leading to the semantic misalignment. In this paper, we propose a new pre-training task named Structural Contrast Pretraining (SCP) to align the structural words in a parallel sentence, enhancing the models’ ability to comprehend cross-lingual representations. Concretely, each structural word in source and target languages is regarded as a positive pair in SCP. Since contrastive learning compares positive and negative pairs, an increase in the frequency of negative pairings could enhance the performance of the resulting model. Therefore, we further propose Cross-lingual Momentum Contrast (CL-MoCo) to increase the number of negative pairs by maintaining a large size of the queue. CL-MoCo extends the original Moco approach into cross-lingual training and jointly optimizes the source-to-target language and target-to-source language representations, resulting in a more suitable encoder for cross-lingual transfer. We conduct extensive experiments to validate the proposed approach on three cross-lingual tasks across five datasets such as MLQA, WikiAnn, etc, and results prove the effectiveness of our method.
%R 10.18653/v1/2023.findings-acl.128
%U https://aclanthology.org/2023.findings-acl.128
%U https://doi.org/10.18653/v1/2023.findings-acl.128
%P 2042-2057
Markdown (Informal)
[Structural Contrastive Pretraining for Cross-Lingual Comprehension](https://aclanthology.org/2023.findings-acl.128) (Chen et al., Findings 2023)
ACL
- Nuo Chen, Linjun Shou, Tengtao Song, Ming Gong, Jian Pei, Jianhui Chang, Daxin Jiang, and Jia Li. 2023. Structural Contrastive Pretraining for Cross-Lingual Comprehension. In Findings of the Association for Computational Linguistics: ACL 2023, pages 2042–2057, Toronto, Canada. Association for Computational Linguistics.