A Sequence Tagging based Framework for Few-Shot Relation Extraction

TL;DR

This paper introduces a novel sequence tagging framework for few-shot relation extraction, enabling relation triple extraction in domains with limited labeled data by adapting joint extraction methods.

Contribution

It defines the few-shot RE task based on sequence tagging and proposes a framework applying two models, Few-shot TPLinker and Few-shot BiTT, demonstrating effectiveness on benchmark datasets.

Findings

Achieved solid results on two few-shot RE tasks

Demonstrated the effectiveness of sequence tagging models in few-shot RE

Provided a new approach for relation extraction in low-resource settings

Abstract

Relation Extraction (RE) refers to extracting the relation triples in the input text. Existing neural work based systems for RE rely heavily on manually labeled training data, but there are still a lot of domains where sufficient labeled data does not exist. Inspired by the distance-based few-shot named entity recognition methods, we put forward the definition of the few-shot RE task based on the sequence tagging joint extraction approaches, and propose a few-shot RE framework for the task. Besides, we apply two actual sequence tagging models to our framework (called Few-shot TPLinker and Few-shot BiTT), and achieves solid results on two few-shot RE tasks constructed from a public dataset.