1
WhoSaidIt: Human-LLM Collaborative Annotation for Text-Based Multilingual Speaker-Attribute Classification
人类与LLM协同注释框架，攻克多语言文本中说话人属性分类任务，效率与质量双提升
arXiv:2605.26070v1 Announce Type: new Abstract: Annotating speaker attributes from text is inherently ambiguous, particularly in multilingual settings…