A Unified Label-Aware Contrastive Learning Framework for Few-Shot Named Entity Recognition

arXiv:2404.17178v1 Announce Type: new
Abstract: Few-shot Named Entity Recognition (NER) aims to extract named entities using only a limited number of labeled examples. Existing contrastive learning methods often suffer from insufficient distinguishability in context vector representation because they either solely rely on label semantics or completely disregard them. To tackle this issue, we propose a unified label-aware token-level contrastive learning framework. Our approach enriches the context by utilizing label semantics as suffix prompts. Additionally, it simultaneously optimizes context-context and context-label …

abstract arxiv context cs.cl examples extract few-shot framework ner recognition representation semantics them type vector