作者位次 #2/6ACL 2026OralBest Social Impact Paper· Natural Language Processing
DIA - HARM : Dialectal Disparities in Harmful Content Detection Across 50 E nglish Dialects
Jason S. Lucas · Matt Murtagh White · Ali Al-Lawati · Uchendu Uchendu · Adaku Uchendu · Dongwon Lee
摘要
Harmful content detectors—particularly disinformation classifiers—are predominantly developed and evaluated on Standard American English (), leaving their robustness to dialectal variation unexplored. We present , the first benchmark for evaluating disinformation detection robustness across 50 En...