Label Drop for Multi-Aspect Relation Modeling in Universal Information Extraction

TL;DR

This paper introduces LDNet, a novel approach for multi-aspect relation modeling in universal information extraction, which reduces decision confusion and irrelevant relation impact, leading to improved performance across diverse tasks and datasets.

Contribution

LDNet's multi-aspect relation modeling and label drop mechanism enhance multi-relation extraction accuracy in UIE, addressing limitations of previous single-target and multi-target instruction methods.

Findings

Outperforms or is competitive with state-of-the-art on 9 tasks and 33 datasets.

Effective in single-modal and multi-modal, few-shot and zero-shot settings.

Reduces decision confusion and irrelevant relation impact.

Abstract

Universal Information Extraction (UIE) has garnered significant attention due to its ability to address model explosion problems effectively. Extractive UIE can achieve strong performance using a relatively small model, making it widely adopted. Extractive UIEs generally rely on task instructions for different tasks, including single-target instructions and multiple-target instructions. Single-target instruction UIE enables the extraction of only one type of relation at a time, limiting its ability to model correlations between relations and thus restricting its capability to extract complex relations. While multiple-target instruction UIE allows for the extraction of multiple relations simultaneously, the inclusion of irrelevant relations introduces decision complexity and impacts extraction accuracy. Therefore, for multi-relation extraction, we propose LDNet, which incorporates multi-aspect relation modeling and a label drop mechanism. By assigning different relations to different levels for understanding and decision-making, we reduce decision confusion. Additionally, the label drop mechanism effectively mitigates the impact of irrelevant relations. Experiments show that LDNet outperforms or achieves competitive performance with state-of-the-art systems on 9 tasks, 33 datasets, in both single-modal and multi-modal, few-shot and zero-shot settings.\footnote{https://github.com/Lu-Yang666/LDNet}