Lost in Pronunciation: Detecting Chinese Offensive Language Disguised by Phonetic Cloaking Replacement
This addresses content moderation challenges for Chinese social media platforms by providing a realistic benchmark and mitigation technique, though it is incremental as it builds on existing methods.
The study tackled the problem of detecting Chinese offensive language disguised by phonetic cloaking replacement, compiling a dataset of 500 real posts and showing that state-of-the-art LLMs achieve only an F1-score of 0.672, with a Pinyin-based prompting strategy recovering much of the lost accuracy.
Phonetic Cloaking Replacement (PCR), defined as the deliberate use of homophonic or near-homophonic variants to hide toxic intent, has become a major obstacle to Chinese content moderation. While this problem is well-recognized, existing evaluations predominantly rely on rule-based, synthetic perturbations that ignore the creativity of real users. We organize PCR into a four-way surface-form taxonomy and compile \ours, a dataset of 500 naturally occurring, phonetically cloaked offensive posts gathered from the RedNote platform. Benchmarking state-of-the-art LLMs on this dataset exposes a serious weakness: the best model reaches only an F1-score of 0.672, and zero-shot chain-of-thought prompting pushes performance even lower. Guided by error analysis, we revisit a Pinyin-based prompting strategy that earlier studies judged ineffective and show that it recovers much of the lost accuracy. This study offers the first comprehensive taxonomy of Chinese PCR, a realistic benchmark that reveals current detectors' limits, and a lightweight mitigation technique that advances research on robust toxicity detection.