A Survey of Graph Neural Network Methods for Relation Extraction
Di Liu, Yungui Zhang, Zhuoqing Li · 2022 IEEE 10th Joint International Information Technology and Artificial Intelligence Conference (ITAIC) · 2022
Relation extraction, as an important part of knowl-edge graph and natural language processing, aims to extract semantic relations between entities by understanding text, which has attracted great interest of researchers. Recently, with the rise of the advanced technology of graph neural networks, numer-ous methods have emerged that employ graph neural network techniques to address relation extraction tasks. However, to the best of our knowledge, few studies provide a complete picture of how and to what extent graph neural networks has been applied to this problem. In this paper, we provide a thorough review of graph neural network-based relation extraction methods. We first present a brief introduction of graph neural networks. Besides, we describe the relation extraction task, comprehensively review the corresponding datasets, and discuss the reason why graph neural networks are adopted for relation extraction tasks. After that, we focus on relation extraction existing approaches built upon graph neural networks, including specialized architectures designed for the basic task as well as novel graph neural network models designed to tackle the challenging problems. We also discuss relation extraction methods based on graph neural networks in combination with pre-trained language models. Finally, we give out our conclusion and provide several further research directions for this research topic.