Consistency Guided Knowledge Retrieval and Denoising in LLMs for Zero-shot Document-level Relation Triplet Extraction

TL;DR

This paper introduces ZeroDocRTE, a zero-shot framework that uses retrieval and denoising of knowledge from LLMs to generate labeled data for document-level relation extraction, reducing reliance on manual annotations.

Contribution

It proposes a novel zero-shot data generation method using LLMs with a chain-of-retrieval prompt and a denoising strategy, enabling effective fine-tuning for relation extraction.

Findings

Outperforms strong baselines on two datasets

Effective zero-shot relation triplet extraction

Improved data quality through denoising

Abstract

Document-level Relation Triplet Extraction (DocRTE) is a fundamental task in information systems that aims to simultaneously extract entities with semantic relations from a document. Existing methods heavily rely on a substantial amount of fully labeled data. However, collecting and annotating data for newly emerging relations is time-consuming and labor-intensive. Recent advanced Large Language Models (LLMs), such as ChatGPT and LLaMA, exhibit impressive long-text generation capabilities, inspiring us to explore an alternative approach for obtaining auto-labeled documents with new relations. In this paper, we propose a Zero-shot Document-level Relation Triplet Extraction (ZeroDocRTE) framework, which generates labeled data by retrieval and denoising knowledge from LLMs, called GenRDK. Specifically, we propose a chain-of-retrieval prompt to guide ChatGPT to generate labeled long-text data step by step. To improve the quality of synthetic data, we propose a denoising strategy based on the consistency of cross-document knowledge. Leveraging our denoised synthetic data, we proceed to fine-tune the LLaMA2-13B-Chat for extracting document-level relation triplets. We perform experiments for both zero-shot document-level relation and triplet extraction on two public datasets. The experimental results illustrate that our GenRDK framework outperforms strong baselines.