A Hierarchical Entity Graph Convolutional Network for Relation Extraction across Documents

TL;DR

This paper introduces a hierarchical entity graph convolutional network for cross-document relation extraction, creating a new dataset for two-hop relations and improving extraction performance over existing sentence-level datasets.

Contribution

It proposes a novel hierarchical GCN model and a new dataset for two-hop cross-document relation extraction, extending beyond sentence-level approaches.

Findings

The dataset covers more relations than existing sentence-level datasets.

The proposed HEGCN model improves F1 score by 1.1% over strong baselines.

Demonstrates the effectiveness of hierarchical GCNs for cross-document relation extraction.

Abstract

Distantly supervised datasets for relation extraction mostly focus on sentence-level extraction, and they cover very few relations. In this work, we propose cross-document relation extraction, where the two entities of a relation tuple appear in two different documents that are connected via a chain of common entities. Following this idea, we create a dataset for two-hop relation extraction, where each chain contains exactly two documents. Our proposed dataset covers a higher number of relations than the publicly available sentence-level datasets. We also propose a hierarchical entity graph convolutional network (HEGCN) model for this task that improves performance by 1.1\% F1 score on our two-hop relation extraction dataset, compared to some strong neural baselines.