UDAA: An Unsupervised Domain Adaptation Adversarial Learning Framework for Zero-Resource Cross-Domain Named Entity Recognition

Baofeng Li, Jianguo Tang, Yu Qin, Yuelou Xu, Yan Lu, Kai Wang, Lei Li, Yanquan Zhou

Abstract: The zero-resource cross-domain named entity recognition (NER) task aims to perform NER in a specific domain where labeled data is unavailable. Existing methods primarily focus on transferring NER knowledge from high-resource to zero-resource domains. However, the challenge lies in effectively transferring NER knowledge between domains due to the inherent differences in entity structures across domains. To tackle this challenge, we propose an Unsupervised Domain Adaptation Adversarial (UDAA) framework, which combines the masked language model auxiliary task with the domain adaptive adversarial network to mitigate inter-domain differences and efficiently facilitate knowledge transfer. Experimental results on CBS, Twitter, and WNUT2016 three datasets demonstrate the effectiveness of our framework. Notably, we achieved new state-of-the-art performance on the three datasets. Our code will be released.