.png){kind=icon}
返回
TS-align: A temporal similarity-aware entity alignment model for temporal knowledge graphs
.png){kind=icon}
PRE
AI
DOI：10.1016/j.inffus.2024.102581.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
Entity Alignment (EA) is a crucial step in knowledge graph fusion, aiming to match equivalent entity pairs across different knowledge graphs (KGs). In recent years, Temporal Knowledge Graphs (TKGs) have extended static KGs by introducing timestamps, providing a new perspective for EA. However, while temporal information is critical in TKGs, the EA approach to existing TKG does not fully utilize this critical resource. To solve this problem, we propose a temporal similarity-aware EA model for TKGs (TS-Align). The model innovatively uses the overlap degree of multiple temporal points and temporal intervals of entities to calculate the temporal similarity matrix, and deeply digs and integrates the temporal and relational information in TKGs through a well-designed encoder- decoder architecture. To verify the effectiveness of TS-Align, we conducted experiments on four standard datasets and compared it to other advanced methods. The experimental results show that the accuracy of TS- Align on EA tasks is significantly improved compared with other methods. Specifically, TS-Align is about 0.4 -1.8 % higher on the Hit@1 metric compared to the most advanced time-aware models. This is about 7.4 -17.6 % higher than the most advanced GCN-based models. This is about 20.8 -42.3 % higher than the most advanced translation-based models. In addition, through ablation experiments, we verify that the proposed temporal similarity matrix, relational embedding, and entity embedding contribute to the significant improvement in performance.
Keyword:
Entity alignment
Knowledge graph fusion
Temporal information
Temporal knowledge graph
Temporal similarity matrix
Unsupervised learning
期刊
IF:
15.5
论文数:
4.2K
被引数:
2.7W
机构
引用论文
MMIEA: Multi-modal Interaction Entity Alignment model for knowledge graphsMMIEA: 知识图谱的多模态交互实体对齐模型.png)
INFORMATION FUSION
INFORMATION FUSIONIF15.5
PREAI
分享收藏TigeCMN: On exploration of temporal interaction graph embedding via Coupled Memory Neural NetworksNEURAL NETWORKS
NEURAL NETWORKSIF6.3
PREAI
分享收藏分享收藏MvTuckER: Multi-view knowledge graphs representation learning based on tensor tucker modelMvTuckER: 基于张量tucker模型的多视图知识图谱表示学习INFORMATION FUSION
INFORMATION FUSIONIF15.5
PREAI
分享收藏FuAlign: Cross-lingual entity alignment via multi-view representation learning of fused knowledge graphsFuAlign: 通过融合知识图的多视图表示学习实现跨语言实体对齐INFORMATION FUSION
OAAI
INFORMATION FUSIONIF15.5
OAAI分享收藏Multi-modal knowledge graphs representation learning via multi-headed self-attention基于多头自我注意的多模态知识图谱表示学习INFORMATION FUSION
INFORMATION FUSIONIF15.5
PREAI
分享收藏DFMKE: A dual fusion multi-modal knowledge graph embedding framework for entity alignmentDFMKE: 一种面向实体对齐的双融合多模态知识图谱嵌入框架INFORMATION FUSION
INFORMATION FUSIONIF15.5
PREAI
分享收藏An effective knowledge graph entity alignment model based on multiple information基于多信息的有效知识图谱实体对齐模型NEURAL NETWORKS
NEURAL NETWORKSIF6.3
PREAI
分享收藏
查看更多内容