AutoTriggER: Label-Efficient and Robust Named Entity Recognition with Auxiliary Trigger Extraction

TL;DR

AutoTriggER introduces an innovative framework that automatically generates entity triggers to enhance label-efficient and robust named entity recognition, outperforming baseline models by leveraging auxiliary cues without costly annotations.

Contribution

The paper presents a novel two-stage framework that automatically generates and uses entity triggers for improved NER performance, reducing reliance on extensive labeled data.

Findings

AutoTriggER outperforms baseline models by nearly 0.5 F1 points.

The framework demonstrates strong label-efficiency and generalization to unseen entities.

Experiments on three datasets validate the effectiveness of trigger-based guidance.

Abstract

Deep neural models for named entity recognition (NER) have shown impressive results in overcoming label scarcity and generalizing to unseen entities by leveraging distant supervision and auxiliary information such as explanations. However, the costs of acquiring such additional information are generally prohibitive. In this paper, we present a novel two-stage framework (AutoTriggER) to improve NER performance by automatically generating and leveraging ``entity triggers'' which are human-readable cues in the text that help guide the model to make better decisions. Our framework leverages post-hoc explanation to generate rationales and strengthens a model's prior knowledge using an embedding interpolation technique. This approach allows models to exploit triggers to infer entity boundaries and types instead of solely memorizing the entity words themselves. Through experiments on three well-studied NER datasets, AutoTriggER shows strong label-efficiency, is capable of generalizing to unseen entities, and outperforms the RoBERTa-CRF baseline by nearly 0.5 F1 points on average.