Textless, Tiny, and Private: Hashed n-Grams with LLM-Teacher Distillation and Local Differential Privacy for Mental-Health Screening on Social Media (CCF C)
Published in IEEE Transactions on Computational Social Systems, 2026
A privacy-preserving approach for mental-health screening on social media using hashed n-grams, LLM-teacher distillation, and local differential privacy.
Keywords: Mental health screening, Local differential privacy, LLM distillation, Social media
Recommended citation: Liu, D., Fan, G., Guo, Q., Zhang, R., & Pan, L. (2026). Textless, Tiny, and Private: Hashed n-Grams with LLM-Teacher Distillation and Local Differential Privacy for Mental-Health Screening on Social Media. IEEE Transactions on Computational Social Systems (Early Access). https://doi.org/10.1109/TCSS.2026.3693692
Download Paper