C2TU method unveils cloaked toxicity in Chinese text
October 12, 2026
The C2TU framework identifies toxic content disguised via homophones in Chinese text using a training-free approach based on substring matching and toxic lexicons.
HOW THIS AFFECTS YOU
●
researcherYou can implement training-free methods to improve content moderation for linguistically complex datasets.