Chinese adversarial text generation method based on punctuation insertion

The susceptibility of natural language processing models to adversarial texts has been a significant concern. Current methods for generating adversarial texts in Chinese were mainly based on replacing characters with visually similar or homophonic ones. However, when faced with robust pre-trained mo...

Full description

Saved in:
Bibliographic Details
Main Authors: ZHANG Qian, YAN Qiao
Format: Article
Language:English
Published: POSTS&TELECOM PRESS Co., LTD 2025-04-01
Series:网络与信息安全学报
Subjects:
Online Access:http://www.cjnis.com.cn/thesisDetails#10.11959/j.issn.2096-109x.2025026
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850132876792692736
author ZHANG Qian
YAN Qiao
author_facet ZHANG Qian
YAN Qiao
author_sort ZHANG Qian
collection DOAJ
description The susceptibility of natural language processing models to adversarial texts has been a significant concern. Current methods for generating adversarial texts in Chinese were mainly based on replacing characters with visually similar or homophonic ones. However, when faced with robust pre-trained models, these methods led to increased perturbations in adversarial texts, resulting in reduced fluency and readability, and thus generating low-quality adversarial texts. Moreover, symbol insertion methods used in English adversarial texts were not entirely applicable to Chinese. Additionally, in a black-box scenario, the lack of prior knowledge made it difficult to generate high-quality adversarial texts. A punctuation-based method for generating adversarial texts for Chinese text classification tasks was proposed. Under a black-box setting, a novel part-of-speech importance calculation was utilized and combined with punctuation insertion to design a character-level perturbation approach suitable for Chinese, achieving the generation of adversarial texts. Experiments were conducted, and the results demonstrated that for text classification tasks, the proposed method significantly improved the attack success rate on LSTM and BERT models trained with two real-world datasets. Furthermore, the method successfully avoided direct destruction of the original sentences and maintained the original meaning. In the tests, a semantic similarity of up to 97% was achieved, which was significantly better than the baseline methods.
format Article
id doaj-art-cb6d97c9929a48c68f69e95a05442bad
institution OA Journals
issn 2096-109X
language English
publishDate 2025-04-01
publisher POSTS&TELECOM PRESS Co., LTD
record_format Article
series 网络与信息安全学报
spelling doaj-art-cb6d97c9929a48c68f69e95a05442bad2025-08-20T02:32:07ZengPOSTS&TELECOM PRESS Co., LTD网络与信息安全学报2096-109X2025-04-011116117499195861Chinese adversarial text generation method based on punctuation insertionZHANG QianYAN QiaoThe susceptibility of natural language processing models to adversarial texts has been a significant concern. Current methods for generating adversarial texts in Chinese were mainly based on replacing characters with visually similar or homophonic ones. However, when faced with robust pre-trained models, these methods led to increased perturbations in adversarial texts, resulting in reduced fluency and readability, and thus generating low-quality adversarial texts. Moreover, symbol insertion methods used in English adversarial texts were not entirely applicable to Chinese. Additionally, in a black-box scenario, the lack of prior knowledge made it difficult to generate high-quality adversarial texts. A punctuation-based method for generating adversarial texts for Chinese text classification tasks was proposed. Under a black-box setting, a novel part-of-speech importance calculation was utilized and combined with punctuation insertion to design a character-level perturbation approach suitable for Chinese, achieving the generation of adversarial texts. Experiments were conducted, and the results demonstrated that for text classification tasks, the proposed method significantly improved the attack success rate on LSTM and BERT models trained with two real-world datasets. Furthermore, the method successfully avoided direct destruction of the original sentences and maintained the original meaning. In the tests, a semantic similarity of up to 97% was achieved, which was significantly better than the baseline methods.http://www.cjnis.com.cn/thesisDetails#10.11959/j.issn.2096-109x.2025026Chinese text classificationadversarial text generationblack-box attack
spellingShingle ZHANG Qian
YAN Qiao
Chinese adversarial text generation method based on punctuation insertion
网络与信息安全学报
Chinese text classification
adversarial text generation
black-box attack
title Chinese adversarial text generation method based on punctuation insertion
title_full Chinese adversarial text generation method based on punctuation insertion
title_fullStr Chinese adversarial text generation method based on punctuation insertion
title_full_unstemmed Chinese adversarial text generation method based on punctuation insertion
title_short Chinese adversarial text generation method based on punctuation insertion
title_sort chinese adversarial text generation method based on punctuation insertion
topic Chinese text classification
adversarial text generation
black-box attack
url http://www.cjnis.com.cn/thesisDetails#10.11959/j.issn.2096-109x.2025026
work_keys_str_mv AT zhangqian chineseadversarialtextgenerationmethodbasedonpunctuationinsertion
AT yanqiao chineseadversarialtextgenerationmethodbasedonpunctuationinsertion