Improving Distantly-Supervised Named Entity Recognition with Self-Collaborative Denoising Learning

TL;DR

This paper introduces Self-Collaborative Denoising Learning (SCDL), a novel framework for improving distantly-supervised NER by jointly training two networks to refine noisy labels, leading to superior results on multiple datasets.

Contribution

The paper proposes a new joint training paradigm with two networks that collaboratively and self-denoise, effectively handling multiple types of label noise in DS-NER.

Findings

SCDL outperforms existing denoising methods on five datasets.

The collaborative approach effectively refines noisy labels.

Experimental results demonstrate significant improvements in NER accuracy.

Abstract

Distantly supervised named entity recognition (DS-NER) efficiently reduces labor costs but meanwhile intrinsically suffers from the label noise due to the strong assumption of distant supervision. Typically, the wrongly labeled instances comprise numbers of incomplete and inaccurate annotation noise, while most prior denoising works are only concerned with one kind of noise and fail to fully explore useful information in the whole training set. To address this issue, we propose a robust learning paradigm named Self-Collaborative Denoising Learning (SCDL), which jointly trains two teacher-student networks in a mutually-beneficial manner to iteratively perform noisy label refinery. Each network is designed to exploit reliable labels via self denoising, and two networks communicate with each other to explore unreliable annotations by collaborative denoising. Extensive experimental results on five real-world datasets demonstrate that SCDL is superior to state-of-the-art DS-NER denoising methods.