Human Label Variation as Stable Signal: Learning Annotator-Specific Explanation Behavior via Cross-Annotator Preference Optimization 事件

Human Label Variation as Stable Signal: Learning Annotator-Specific Explanation Behavior via Cross-Annotator Preference Optimization arXiv:2605.28802v1 Announce Type: new Abstract: Free-text explanations extend human label variation (HLV) beyond label disagreement by revealing the reasoning and preferences behind annotators' decisions. We study whether large language models (LLMs) can learn and reproduce such annotator-specific label-explanation behavior. Using two sentence-pair tasks with four