Relational Knowledge Prediction via Dynamic Bi-Mode Embedding

Knowledge graphs are a crucial concept in artificial intelligence with a wide spectrum of real-life applications. Nonetheless, they are currently suffering from the incompleteness issue, i.e., relational knowledge in the graphs may not yet meet the practical needs. To address this issue, mainstream...

Full description

Saved in:
Bibliographic Details
Main Authors: Yang Fang, Xiang Zhao, Zhen Tan, Weidong Xiao
Format: Article
Language:English
Published: IEEE 2018-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/8353191/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849236296193540096
author Yang Fang
Xiang Zhao
Zhen Tan
Weidong Xiao
author_facet Yang Fang
Xiang Zhao
Zhen Tan
Weidong Xiao
author_sort Yang Fang
collection DOAJ
description Knowledge graphs are a crucial concept in artificial intelligence with a wide spectrum of real-life applications. Nonetheless, they are currently suffering from the incompleteness issue, i.e., relational knowledge in the graphs may not yet meet the practical needs. To address this issue, mainstream solutions propose to predict links by using compositional models or translation models. However, the prediction accuracy is still of particular concern. In this paper, we propose a new method, namely, <italic>Bi-Mult</italic>, which combines the advantages of compositional models and translation models. <italic>Bi-Mult</italic> is based on the compositional model, such that an entity (resp. relation) embedding is decomposed into two parts, one is to represent intra-entity (resp. relation) state and the other is for inter-entity (resp. relation) state, and we call such embedding as bi-mode embedding. In addition, the bi-mode relation embedding enhances relation&#x2019;s interaction with entities, resulting its improvement on handling antisymmetric relations. Moreover, we incorporate mapping matrices in translation models through bi-mode entity embedding to construct dynamic embeddings for expressing complex relations, such as &#x201C;1-to-N&#x201D;, &#x201C;N-to-1,&#x201D; and &#x201C;N-to-N&#x201D; relations. In experiments, we evaluate our method on the benchmark data sets and the task of link prediction, and our method is demonstrated to outperform state-of-the-art methods consistently and significantly.
format Article
id doaj-art-a07c29d7ddb1441b95a59d21a9f811e9
institution Kabale University
issn 2169-3536
language English
publishDate 2018-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-a07c29d7ddb1441b95a59d21a9f811e92025-08-20T04:02:18ZengIEEEIEEE Access2169-35362018-01-016257152572310.1109/ACCESS.2018.28321658353191Relational Knowledge Prediction via Dynamic Bi-Mode EmbeddingYang Fang0Xiang Zhao1https://orcid.org/0000-0001-6339-0219Zhen Tan2Weidong Xiao3Science and Technology on Information Systems Engineering Laboratory, National University of Defense Technology, Changsha, ChinaScience and Technology on Information Systems Engineering Laboratory, National University of Defense Technology, Changsha, ChinaScience and Technology on Information Systems Engineering Laboratory, National University of Defense Technology, Changsha, ChinaScience and Technology on Information Systems Engineering Laboratory, National University of Defense Technology, Changsha, ChinaKnowledge graphs are a crucial concept in artificial intelligence with a wide spectrum of real-life applications. Nonetheless, they are currently suffering from the incompleteness issue, i.e., relational knowledge in the graphs may not yet meet the practical needs. To address this issue, mainstream solutions propose to predict links by using compositional models or translation models. However, the prediction accuracy is still of particular concern. In this paper, we propose a new method, namely, <italic>Bi-Mult</italic>, which combines the advantages of compositional models and translation models. <italic>Bi-Mult</italic> is based on the compositional model, such that an entity (resp. relation) embedding is decomposed into two parts, one is to represent intra-entity (resp. relation) state and the other is for inter-entity (resp. relation) state, and we call such embedding as bi-mode embedding. In addition, the bi-mode relation embedding enhances relation&#x2019;s interaction with entities, resulting its improvement on handling antisymmetric relations. Moreover, we incorporate mapping matrices in translation models through bi-mode entity embedding to construct dynamic embeddings for expressing complex relations, such as &#x201C;1-to-N&#x201D;, &#x201C;N-to-1,&#x201D; and &#x201C;N-to-N&#x201D; relations. In experiments, we evaluate our method on the benchmark data sets and the task of link prediction, and our method is demonstrated to outperform state-of-the-art methods consistently and significantly.https://ieeexplore.ieee.org/document/8353191/Knowledge embeddingrepresentation learning
spellingShingle Yang Fang
Xiang Zhao
Zhen Tan
Weidong Xiao
Relational Knowledge Prediction via Dynamic Bi-Mode Embedding
IEEE Access
Knowledge embedding
representation learning
title Relational Knowledge Prediction via Dynamic Bi-Mode Embedding
title_full Relational Knowledge Prediction via Dynamic Bi-Mode Embedding
title_fullStr Relational Knowledge Prediction via Dynamic Bi-Mode Embedding
title_full_unstemmed Relational Knowledge Prediction via Dynamic Bi-Mode Embedding
title_short Relational Knowledge Prediction via Dynamic Bi-Mode Embedding
title_sort relational knowledge prediction via dynamic bi mode embedding
topic Knowledge embedding
representation learning
url https://ieeexplore.ieee.org/document/8353191/
work_keys_str_mv AT yangfang relationalknowledgepredictionviadynamicbimodeembedding
AT xiangzhao relationalknowledgepredictionviadynamicbimodeembedding
AT zhentan relationalknowledgepredictionviadynamicbimodeembedding
AT weidongxiao relationalknowledgepredictionviadynamicbimodeembedding