SRGCN: Graph-based multi-hop reasoning on knowledge graphs
Wang, Zikang1,2; Li, Linjing1,2,3; Zeng, Daniel1,2,3
Source PublicationNeurocomputing
ISSN0925-2312
2021
Volume454Issue:454Pages:280-290
Corresponding AuthorLi, Linjing(linjing.li@ia.ac.cn)
Abstract

Learning to infer missing links is one of the fundamental tasks in the knowledge graph. Instead of reason- ing based on separate paths in the existing methods, in this paper, we propose a new model, Sequential Relational Graph Convolutional Network (SRGCN), which treats the multiple paths between an entity pair as a sequence of subgraphs. Specifically, to reason the relationship between two entities, we first con- struct a graph for the entities based on the knowledge graph and serialize the graph to a sequence. For each hop in the sequence, Relational Graph Convolutional Network (R-GCN) is then applied to update the embeddings of the entities. The updated embedding of the tail entity contains information of the entire graph, hence the relationship between two entities can be inferred from it. Compared to the exist- ing approaches that deal with paths separately, SRGCN treats the graph as a whole, which can encode structural information and interactions between paths better. Experiments show that SRGCN outper- forms path-based baselines on both link and fact prediction tasks. We also show that SRGCN is highly effi- cient in the sense that only one epoch of training is enough to achieve high accuracy, and even partial datasets can lead to competitive performance.

Keywordknowledge graph multi-hop reasoning graph convolutional network
DOI10.1016/j.neucom.2021.05.016
Indexed BySCI
Language英语
Funding ProjectNational Key Research and Development Program of China[2020AAA0103405] ; National Natural Science Foundation of China[71621002] ; Strategic Priority Research Program of Chinese Academy of Sciences[XDA27030100]
Funding OrganizationNational Key Research and Development Program of China ; National Natural Science Foundation of China ; Strategic Priority Research Program of Chinese Academy of Sciences
WOS Research AreaComputer Science
WOS SubjectComputer Science, Artificial Intelligence
WOS IDWOS:000672468800003
PublisherELSEVIER
Sub direction classification机器学习
Citation statistics
Cited Times:7[WOS]   [WOS Record]     [Related Records in WOS]
Document Type期刊论文
Identifierhttp://ir.ia.ac.cn/handle/173211/44380
Collection多模态人工智能系统全国重点实验室_互联网大数据与信息安全
Affiliation1.State Key Laboratory of Management and Control for Complex Systems, Institute of Automation, Chinese Academy of Sciences
2.School of Artificial Intelligence, University of Chinese Academy of Sciences
3.Shenzhen Artificial Intelligence and Data Science Institute (Longhua)
First Author AffilicationInstitute of Automation, Chinese Academy of Sciences
Recommended Citation
GB/T 7714
Wang, Zikang,Li, Linjing,Zeng, Daniel. SRGCN: Graph-based multi-hop reasoning on knowledge graphs[J]. Neurocomputing,2021,454(454):280-290.
APA Wang, Zikang,Li, Linjing,&Zeng, Daniel.(2021).SRGCN: Graph-based multi-hop reasoning on knowledge graphs.Neurocomputing,454(454),280-290.
MLA Wang, Zikang,et al."SRGCN: Graph-based multi-hop reasoning on knowledge graphs".Neurocomputing 454.454(2021):280-290.
Files in This Item: Download All
File Name/Size DocType Version Access License
2021_neurocomputing_(1540KB)期刊论文作者接受稿开放获取CC BY-NC-SAView Download
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[Wang, Zikang]'s Articles
[Li, Linjing]'s Articles
[Zeng, Daniel]'s Articles
Baidu academic
Similar articles in Baidu academic
[Wang, Zikang]'s Articles
[Li, Linjing]'s Articles
[Zeng, Daniel]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[Wang, Zikang]'s Articles
[Li, Linjing]'s Articles
[Zeng, Daniel]'s Articles
Terms of Use
No data!
Social Bookmark/Share
File name: 2021_neurocomputing_srgcn.pdf
Format: Adobe PDF
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.