Span-based few-shot event detection via aligning external knowledge

Tongtao Ling; Lei Chen; Yutao Lai; Hai-Lin Liu

doi:10.1016/j.neunet.2024.106327

Span-based few-shot event detection via aligning external knowledge

Neural Netw. 2024 Apr 18:176:106327. doi: 10.1016/j.neunet.2024.106327. Online ahead of print.

Authors

Tongtao Ling¹, Lei Chen², Yutao Lai¹, Hai-Lin Liu³

Affiliations

¹ School of Mathematics and Statistics, Guangdong University of Technology, Guangzhou 510520, China.
² School of Mathematics and Statistics, Guangdong University of Technology, Guangzhou 510520, China. Electronic address: chenlei3@gdut.edu.cn.
³ School of Mathematics and Statistics, Guangdong University of Technology, Guangzhou 510520, China. Electronic address: https://github.com/rickltt/FewShot_ED.

PMID: 38692187
DOI: 10.1016/j.neunet.2024.106327

Abstract

Few-shot Event Detection (FSED) aims to identify novel event types in new domains with very limited annotated data. Previous PN-based (Prototypical Network) joint methods suffer from insufficient learning of token-wise label dependency and inaccurate prototypes. To solve these problems, we propose a span-based FSED model, called SpanFSED, which decomposes FSED into two subprocesses, including span extractor and event classifier. In span extraction, we convert sequential labels into a global boundary matrix that enables the span extractor to acquire precise boundary information irrespective of label dependency. In event classification, we align event types with an outside knowledge base like FrameNet and construct an enhanced support set, which injects more trigger information into the prototypical network of event prototypes. The superior performance of SpanFSED is demonstrated through extensive experiments on four event detection datasets, i.e., ACE2005, ERE, MAVEN and FewEvent. Access to our code and data is facilitated through the following link: .

Keywords: External knowledge base; Few-shot event detection; Global boundary matrix; Prototypical network.