3DRTE: 3D Rotation Embedding in Temporal Knowledge Graph
Temporal knowledge graph (TKG) embedding has received increasing attention in the academia. However, most existing methods are extensions of traditional translation models. Due to their intrinsic limitations, it is often difficult for such methods to effectively model essential characteristics of TK...
Main Authors: | , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
IEEE
2020-01-01
|
Series: | IEEE Access |
Subjects: | |
Online Access: | https://ieeexplore.ieee.org/document/9253009/ |
id |
doaj-999f617716d542e39dc71386250a47a3 |
---|---|
record_format |
Article |
spelling |
doaj-999f617716d542e39dc71386250a47a32021-03-30T03:56:44ZengIEEEIEEE Access2169-35362020-01-01820751520752310.1109/ACCESS.2020.303689792530093DRTE: 3D Rotation Embedding in Temporal Knowledge GraphJingbin Wang0https://orcid.org/0000-0003-1321-1491Wang Zhang1https://orcid.org/0000-0002-1610-2429Xinyuan Chen2https://orcid.org/0000-0001-7611-7609Jing Lei3https://orcid.org/0000-0001-9145-7070Xiaolian Lai4https://orcid.org/0000-0002-1926-1433College of Mathematics and Computer Science/College of Software, Fuzhou University, Fuzhou, ChinaCollege of Mathematics and Computer Science/College of Software, Fuzhou University, Fuzhou, ChinaDepartment of Information Engineering, Fuzhou Melbourne Polytechnic, Fuzhou, ChinaCollege of Mathematics and Computer Science/College of Software, Fuzhou University, Fuzhou, ChinaCollege of Mathematics and Computer Science/College of Software, Fuzhou University, Fuzhou, ChinaTemporal knowledge graph (TKG) embedding has received increasing attention in the academia. However, most existing methods are extensions of traditional translation models. Due to their intrinsic limitations, it is often difficult for such methods to effectively model essential characteristics of TKG, namely three basic relation patterns including symmetry/antisymmetry, inversion, and composition. In this paper, a new 3-Dimensional Rotation Temporal Embedding (3DRTE) method is proposed. Firstly, we selectively fuse temporal and relational features of fact triples by taking advantages of self-attention mechanism in processing sequential information. Then, entities are modelled as points in three-dimensional space, and the relations are interpreted as two isoclinic rotations between entities with Quaternion. Experimental results on several public datasets show that our method obtains state-of-the-art results.https://ieeexplore.ieee.org/document/9253009/Temporal knowledge graph3D rotation embeddingself-attentionquaternion |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Jingbin Wang Wang Zhang Xinyuan Chen Jing Lei Xiaolian Lai |
spellingShingle |
Jingbin Wang Wang Zhang Xinyuan Chen Jing Lei Xiaolian Lai 3DRTE: 3D Rotation Embedding in Temporal Knowledge Graph IEEE Access Temporal knowledge graph 3D rotation embedding self-attention quaternion |
author_facet |
Jingbin Wang Wang Zhang Xinyuan Chen Jing Lei Xiaolian Lai |
author_sort |
Jingbin Wang |
title |
3DRTE: 3D Rotation Embedding in Temporal Knowledge Graph |
title_short |
3DRTE: 3D Rotation Embedding in Temporal Knowledge Graph |
title_full |
3DRTE: 3D Rotation Embedding in Temporal Knowledge Graph |
title_fullStr |
3DRTE: 3D Rotation Embedding in Temporal Knowledge Graph |
title_full_unstemmed |
3DRTE: 3D Rotation Embedding in Temporal Knowledge Graph |
title_sort |
3drte: 3d rotation embedding in temporal knowledge graph |
publisher |
IEEE |
series |
IEEE Access |
issn |
2169-3536 |
publishDate |
2020-01-01 |
description |
Temporal knowledge graph (TKG) embedding has received increasing attention in the academia. However, most existing methods are extensions of traditional translation models. Due to their intrinsic limitations, it is often difficult for such methods to effectively model essential characteristics of TKG, namely three basic relation patterns including symmetry/antisymmetry, inversion, and composition. In this paper, a new 3-Dimensional Rotation Temporal Embedding (3DRTE) method is proposed. Firstly, we selectively fuse temporal and relational features of fact triples by taking advantages of self-attention mechanism in processing sequential information. Then, entities are modelled as points in three-dimensional space, and the relations are interpreted as two isoclinic rotations between entities with Quaternion. Experimental results on several public datasets show that our method obtains state-of-the-art results. |
topic |
Temporal knowledge graph 3D rotation embedding self-attention quaternion |
url |
https://ieeexplore.ieee.org/document/9253009/ |
work_keys_str_mv |
AT jingbinwang 3drte3drotationembeddingintemporalknowledgegraph AT wangzhang 3drte3drotationembeddingintemporalknowledgegraph AT xinyuanchen 3drte3drotationembeddingintemporalknowledgegraph AT jinglei 3drte3drotationembeddingintemporalknowledgegraph AT xiaolianlai 3drte3drotationembeddingintemporalknowledgegraph |
_version_ |
1724182648687951872 |