Events Beyond ACE: Curated Training for Events

TL;DR

This paper introduces curated training (CT), a human-driven annotation method that uses interactive search to identify informative snippets, leading to improved event extraction performance and more efficient learning compared to traditional annotation methods.

Contribution

The paper presents curated training as a novel annotation approach that enhances event extraction accuracy and efficiency through interactive search and targeted snippet annotation.

Findings

CT reduces error by 6% when combined with ACE.

Learning curves for CT plateau more slowly than full-document annotation.

Expert annotators outperform ACE in less than ninety minutes.

Abstract

We explore a human-driven approach to annotation, curated training (CT), in which annotation is framed as teaching the system by using interactive search to identify informative snippets of text to annotate, unlike traditional approaches which either annotate preselected text or use active learning. A trained annotator performed 80 hours of CT for the thirty event types of the NIST TAC KBP Event Argument Extraction evaluation. Combining this annotation with ACE results in a 6% reduction in error and the learning curve of CT plateaus more slowly than for full-document annotation. 3 NLP researchers performed CT for one event type and showed much sharper learning curves with all three exceeding ACE performance in less than ninety minutes, suggesting that CT can provide further benefits when the annotator deeply understands the system.