MPE3: Learning meta-prompt with entity-enhanced semantics for few-shot named entity recognition

Yuwei Xia, Zhao Tong, Liang Wang, Qiang Liu, Shu Wu, Xiaoyu Zhang

Abstract: Highlights•We explore the semantic correlation between labels and named entities in the NER task.•We construct the prompt with entity-enhanced semantics.•We illustrate the difficulties of effectively training prompts for few-shot scenarios.•We conduct extensive experiments to verify the superiority of MPE3.