Addressing NER Annotation Noises with Uncertainty-Guided Tree-Structured CRFs

Jian Liu, Weichang Liu, Yufeng Chen, Jinan Xu, Zhe Zhao

Submission Type: Regular Long Paper

Submission Track: Information Extraction

Keywords: Named Entity Recognition (NER), Partial and Incorrect Annotation, Uncertainty, constituency tree parsing

Abstract:

Real-world named entity recognition (NER) datasets are notorious for their noisy nature, attributed to annotation errors, inconsistencies, and subjective interpretations. Such noises present a substantial challenge for traditional supervised learning methods.
In this paper, we present a new and unified approach to tackle annotation noises for NER. Our method considers NER as a constituency tree parsing problem, utilizing a tree-structured Conditional Random Fields (CRFs) with uncertainty evaluation for integration.
Through extensive experiments conducted on four real-world datasets, we demonstrate the effectiveness of our model in addressing both partial and incorrect annotation errors. Remarkably, our model exhibits superb performance even in extreme scenarios with 90\% annotation noise.

Submission Number: 2773