Context-aware Adversarial Training for Name Regularity Bias in Named Entity Recognition

TitreContext-aware Adversarial Training for Name Regularity Bias in Named Entity Recognition
Type de publicationJournal Article
Année de publication2021
AuteursGhaddar, A., P. Langlais, A. Rashid, and M. Rezagholizadeh
JournalTransactions of the Association for Computational Linguistics
Volume9
Pagination586-604
Date Published07
ISSN2307-387X
RésuméIn this work, we examine the ability of NER models to use contextual information when predicting the type of an ambiguous entity. We introduce NRB, a new testbed carefully designed to diagnose Name Regularity Bias of NER models. Our results indicate that all state-of-the-art models we tested show such a bias; BERT fine-tuned models significantly outperforming feature-based (LSTM-CRF) ones on NRB, despite having comparable (sometimes lower) performance on standard benchmarks.To mitigate this bias, we propose a novel model-agnostic training method that adds learnable adversarial noise to some entity mentions, thus enforcing models to focus more strongly on the contextual signal, leading to significant gains on NRB. Combining it with two other training strategies, data augmentation and parameter freezing, leads to further gains.
URLhttps://direct.mit.edu/tacl/article/doi/10.1162/tacl_a_00386/
DOI10.1162/tacl_a_00386