Robustness to Capitalization Errors in Named Entity Recognition

TL;DR

This paper introduces a data augmentation method to improve named entity recognition models' robustness to capitalization errors, maintaining performance on well-formed text while enhancing noise resilience.

Contribution

The authors propose a simple data augmentation approach that enables models to adaptively use orthographic cues, improving robustness to capitalization errors without sacrificing accuracy on clean data.

Findings

Enhanced robustness to capitalization errors across models and languages.

Minimal performance loss on well-formed text.

Improved generalization on noisy user-generated text.

Abstract

Robustness to capitalization errors is a highly desirable characteristic of named entity recognizers, yet we find standard models for the task are surprisingly brittle to such noise. Existing methods to improve robustness to the noise completely discard given orthographic information, mwhich significantly degrades their performance on well-formed text. We propose a simple alternative approach based on data augmentation, which allows the model to \emph{learn} to utilize or ignore orthographic information depending on its usefulness in the context. It achieves competitive robustness to capitalization errors while making negligible compromise to its performance on well-formed text and significantly improving generalization power on noisy user-generated text. Our experiments clearly and consistently validate our claim across different types of machine learning models, languages, and dataset sizes.