Neural Architectures for Named Entity Recognition

TL;DR

This paper introduces two neural network architectures for named entity recognition that leverage character-based and unsupervised word representations, achieving state-of-the-art results across multiple languages without language-specific resources.

Contribution

The paper presents two novel neural architectures for NER, combining bidirectional LSTMs with CRFs and a transition-based segment labeling approach, improving performance without language-specific resources.

Findings

Achieved state-of-the-art NER performance in four languages.

Models do not require language-specific knowledge or gazetteers.

Effective use of character-based and unsupervised word representations.

Abstract

State-of-the-art named entity recognition systems rely heavily on hand-crafted features and domain-specific knowledge in order to learn effectively from the small, supervised training corpora that are available. In this paper, we introduce two new neural architectures---one based on bidirectional LSTMs and conditional random fields, and the other that constructs and labels segments using a transition-based approach inspired by shift-reduce parsers. Our models rely on two sources of information about words: character-based word representations learned from the supervised corpus and unsupervised word representations learned from unannotated corpora. Our models obtain state-of-the-art performance in NER in four languages without resorting to any language-specific knowledge or resources such as gazetteers.