Nested Named Entity Recognition as Latent Lexicalized Constituency Parsing

TL;DR

This paper introduces a novel nested NER approach using lexicalized constituency trees with headword annotations, leveraging efficient algorithms and additional training strategies to achieve state-of-the-art results.

Contribution

It proposes a new nested NER model based on lexicalized constituency parsing, incorporating headword information and novel training strategies for improved performance.

Findings

Achieves state-of-the-art results on ACE2004, ACE2005, and NNE datasets.

Demonstrates competitive performance on GENIA dataset.

Maintains fast inference speed.

Abstract

Nested named entity recognition (NER) has been receiving increasing attention. Recently, (Fu et al, 2021) adapt a span-based constituency parser to tackle nested NER. They treat nested entities as partially-observed constituency trees and propose the masked inside algorithm for partial marginalization. However, their method cannot leverage entity heads, which have been shown useful in entity mention detection and entity typing. In this work, we resort to more expressive structures, lexicalized constituency trees in which constituents are annotated by headwords, to model nested entities. We leverage the Eisner-Satta algorithm to perform partial marginalization and inference efficiently. In addition, we propose to use (1) a two-stage strategy (2) a head regularization loss and (3) a head-aware labeling loss in order to enhance the performance. We make a thorough ablation study to investigate the functionality of each component. Experimentally, our method achieves the state-of-the-art performance on ACE2004, ACE2005 and NNE, and competitive performance on GENIA, and meanwhile has a fast inference speed.