SaGCN: Structure-Aware Graph Convolution Network for Document-Level Relation Extraction

  • Shuangji Yang
  • , Taolin Zhang
  • , Danning Su
  • , Nan Hu
  • , Wei Nong
  • , Xiaofeng He*
  • *Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

1 Scopus citations

Abstract

Document-level Relation Extraction(DocRE) aims at extracting semantic relations among entities in documents. However, current models lack long-range dependency information and the reasoning ability to extract essential structure information from the text. In this paper, we propose SaGCN, a Structure-aware Graph Convolution Network, extracting relation with explicit and implicit dependency structure. Specifically, we generate the implicit graph by sampling from a discrete and continuous distribution, then dynamically fuse the implicit soft structure with the dependent hard structure. Experimental results of SaGCN outperform the performance achieved by current state-of-the-art various baseline models on the DocRED dataset.

Original languageEnglish
Title of host publicationAdvances in Knowledge Discovery and Data Mining - 25th Pacific-Asia Conference, PAKDD 2021, Proceedings
EditorsKamal Karlapalem, Hong Cheng, Naren Ramakrishnan, R. K. Agrawal, P. Krishna Reddy, Jaideep Srivastava, Tanmoy Chakraborty
PublisherSpringer Science and Business Media Deutschland GmbH
Pages377-389
Number of pages13
ISBN (Print)9783030757670
DOIs
StatePublished - 2021
Event25th Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2021 - Virtual, Online
Duration: 11 May 202114 May 2021

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume12714 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference25th Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2021
CityVirtual, Online
Period11/05/2114/05/21

Keywords

  • Document-level relation extraction
  • Graph neural network
  • Structure-aware information injection

Fingerprint

Dive into the research topics of 'SaGCN: Structure-Aware Graph Convolution Network for Document-Level Relation Extraction'. Together they form a unique fingerprint.

Cite this