Lost in Pronunciation: Detecting Chinese Offensive Language Disguised by Phonetic Cloaking Replacement

Challenge:	Phonetic Cloaking Replacement (PCR) is a problem in content moderation in China.
Approach:	They organize PCR into a four-way surface-form taxonomy and compile PCR-ToxiCN, a dataset of 500 phonetically cloaked offensive posts gathered from the RedNote platform.
Outcome:	The proposed model achieves only an F1-score and zero-shot chain-of-thought prompting pushes performance even lower.

Similar Papers

Overview

Challenge

Approach

Outcome

ToxiCloakCN: Evaluating Robustness of Offensive Language Detection in Chinese with Cloaking Perturbations (2024.emnlp-main)

Copied to clipboard

Yunze Xiao, Yujia Hu, Kenny Choo, Roy Lee

Responsible & Trustworthy NLP

Challenge:	Existing large language models struggle with systematically perturbed data designed to evade detection mechanisms.
Approach:	They propose a large language model with homophonic substitutions and emoji transformations to test their models' robustness against cloaking perturbations.
Outcome:	The proposed model underperforms in detecting offensive content when perturbations are applied to Chinese language datasets.

Enhancing Chinese Offensive Language Detection with Homophonic Perturbation (2025.emnlp-main)

Copied to clipboard

Junqi Wu, Shujie Ji, Kang Zhong, Huiling Peng, null Zhendongxiao, Xiongding Liu, Wu Wei

Responsible & Trustworthy NLP

Challenge:	Detecting offensive language in Chinese is challenging due to homophonic substitutions used to evade detection.
Approach:	They propose to use HED-COLD to build a large-scale homophonic dataset for Chinese offensive language detection and a homophone-aware pretraining strategy to learn phonetics and orthography.
Outcome:	The proposed framework achieves state-of-the-art performance on the COLD test set and the toxicity benchmark ToxiCloakCN.

Fortifying Toxic Speech Detectors Against Veiled Toxicity (2020.emnlp-main)

Copied to clipboard

Xiaochuang Han, Yulia Tsvetkov

Responsible & Trustworthy NLP

Challenge:	Modern toxic speech detectors are incompetent in recognizing disguised offensive language, such as adversarial attacks that deliberately avoid known toxic lexicons.
Approach:	They propose a framework that fortifies existing toxic speech detectors without a large labeled corpus of veiled toxicity.
Outcome:	The proposed framework is aimed at fortifying existing toxic speech detectors without a large labeled corpus of disguised offensive language.

Chinese Toxic Language Mitigation via Sentiment Polarity Consistent Rewrites (2025.emnlp-main)

Copied to clipboard

Xintong Wang, Yixiao Liu, Jingheng Pan, Liang Ding, Longyue Wang, Chris Biemann

Sentiment Analysis

Challenge:	Large language models (LLMs) can be effective at rewriting toxic content, but they often default to overly polite rewrites, distorting the emotional tone and communicative intent.
Approach:	They evaluate 17 large language models with variant architectures to evaluate their ability to rewrite toxic content while preserving the speaker's original intent.
Outcome:	The first Chinese detoxification dataset explicitly designed to preserve sentiment polarity is evaluated across five real-world scenarios.

Exploring Multimodal Challenges in Toxic Chinese Detection: Taxonomy, Benchmark, and Findings (2025.findings-acl)

Copied to clipboard

Shujian Yang, Shiyao Cui, Chuanrui Hu, Haicheng Wang, Tianwei Zhang, Minlie Huang, Jialiang Lu, Han Qiu

Language Models

Challenge:	Recent studies show that character substitutions in toxic Chinese text can confuse state-of-the-art LLMs.
Approach:	They propose a taxonomy of 3 perturbation strategies and 8 specific approaches in Chinese text to assess if they can detect perturbed Chinese toxic contents.
Outcome:	The proposed model can detect perturbed Chinese text with 8 different approaches . the proposed model is compared with 9 other LLMs from the US and China .

Facilitating Fine-grained Detection of Chinese Toxic Language: Hierarchical Taxonomy, Resources, and Benchmarks (2023.acl-long)

Copied to clipboard

Junyu Lu, Bo Xu, Xiaokun Zhang, Changrong Min, Liang Yang, Hongfei Lin

Challenge:	Existing datasets suffer from a lack of fine-grained annotations, such as the toxic type and expressions with indirect toxicity.
Approach:	They propose a benchmark model to detect toxic language by incorporating lexical features into a Chinese dataset to facilitate fine-grained annotations.
Outcome:	The proposed model is based on insulting vocabulary containing implicit profanity and is able to detect toxic language with lexical features.

Penetrating Linguistic Disguises: A Slang-aware Label-Aligned Framework for Fine-Grained Toxicity Extraction in Chinese Hate Speech Detection (2026.findings-acl)

Copied to clipboard

Wei Liu, Xiaoliang Chen, Duoqian Miao, Xu Gu, Xianyong Li, Yajun Du

Information Extraction & Text Mining

Challenge:	Flexible word boundaries and linguistic obfuscation, particularly slang, challenge precise span-level hate speech detection in Chinese.
Approach:	They propose a Slang-aware Label-Aligned Framework that maps slang to explicit hate semantics and uses task-specific branches to mitigate feature interference.
Outcome:	The proposed framework reduces ambiguity by mapping obscure slang to explicit hate semantics.

Realistic Evaluation of Toxicity in Large Language Models (2024.findings-acl)

Copied to clipboard

Tinh Luong, Thanh-Thien Le, Linh Ngo, Thien Nguyen

Semantic Text Processing

Challenge:	a large amount of data exposes large language models to toxicity and bias . prompt engineering can be easily bypassed with minimal prompt engineering.
Approach:	They propose a dataset that uses manually crafted prompts to nullify protective layers of large language models.
Outcome:	The proposed dataset shows that prompts can nullify protective layers of large language models.

FanChuan: A Multilingual and Graph-Structured Benchmark For Parody Detection and Analysis (2025.findings-acl)

Copied to clipboard

Yilun Zheng, Sha Li, Fangkun Wu, Yang Ziyi, Lin Hongchao, Zhichao Hu, Cai Xinjun, Ziming Wang, Jinxuan Chen, Sitao Luan, Jiahao Xu, Lihui Chen

Multimodality

Challenge:	Parody is an emerging phenomenon on social media, where individuals imitate a role or position opposite to their own . limited available data and deficient diversity in current datasets hinder study of parody .
Approach:	They build a dataset of parody users and annotated comments from both English and Chinese corpora to test parody detection and comment sentiment analysis.
Outcome:	The proposed datasets provide richer contextual information, which is lacking in existing datasets.

Offensive Content Detection via Synthetic Code-Switched Text (2022.coling-1)

Copied to clipboard

Cesa Salaam, Franck Dernoncourt, Trung Bui, Danda Rawat, Seunghyun Yoon

Text Classification

Challenge:	Existing methods to detect offensive content in social media platforms are limited by the availability of labeled code-switched data.
Approach:	They propose a method for generating synthetic code-switched offensive content data using human-generated data and a keyword classification baseline.
Outcome:	The proposed algorithm can be used to generate synthetic code-switched offensive content data and train it on human-generated data.