cross-platform-generalization

This paper proposes a Cross-Platform Fairness Evaluation framework to audit transformer models for mental health NLP, revealing significant performance and calibration failures when models are applied across different social media platforms.