Local Additivity Based Data Augmentation for Semi-supervised NER

TL;DR

This paper introduces LADA, a novel data augmentation technique for semi-supervised NER that interpolates between samples to generate virtual data, improving entity recognition with minimal labeled data.

Contribution

The paper proposes a new Local Additivity based Data Augmentation (LADA) method with intra- and inter-sentence interpolation for semi-supervised NER, extending it with a consistency loss for unlabeled data.

Findings

LADA outperforms several strong baselines on NER benchmarks.

Interpolation-based augmentation enhances entity and context learning.

The method effectively reduces reliance on labeled data.

Abstract

Named Entity Recognition (NER) is one of the first stages in deep language understanding yet current NER models heavily rely on human-annotated data. In this work, to alleviate the dependence on labeled data, we propose a Local Additivity based Data Augmentation (LADA) method for semi-supervised NER, in which we create virtual samples by interpolating sequences close to each other. Our approach has two variations: Intra-LADA and Inter-LADA, where Intra-LADA performs interpolations among tokens within one sentence, and Inter-LADA samples different sentences to interpolate. Through linear additions between sampled training data, LADA creates an infinite amount of labeled data and improves both entity and context learning. We further extend LADA to the semi-supervised setting by designing a novel consistency loss for unlabeled data. Experiments conducted on two NER benchmarks demonstrate the effectiveness of our methods over several strong baselines. We have publicly released our code at https://github.com/GT-SALT/LADA.