ConvD: Attention enhanced dynamic convolutional embeddings for knowledge graph completion
Author Identifier (ORCID)
Jianxin Li: https://orcid.org/0000-0002-9059-330X
Abstract
Knowledge graphs often suffer from incompleteness issues, which can be alleviated through information completion. However, current state-of-the-art deep knowledge convolutional embedding models rely on external convolution kernels and conventional convolution processes, which limits the feature interaction capability of the model. This paper introduces a novel dynamic convolutional embedding model, named ConvD, which directly reshapes relation embeddings into multiple internal convolution kernels. This approach effectively enhances the feature interactions between relation embeddings and entity embeddings. Simultaneously, we incorporate a priori knowledgeoptimized attention mechanism that assigns distinct contribution weights to multiple relational convolution kernels during dynamic convolution, further boosting the expressive power of the model. Extensive experiments on various datasets show that our proposed model consistently outperforms the state-of-the-art baseline methods, with average improvements ranging from 3.28% to 14.69% across all the evaluation metrics, while the number of parameters is reduced by 50.66% to 85.40% compared to other state-of-the-art models.
Document Type
Journal Article
Date of Publication
1-1-2025
Volume
37
Issue
9
Publisher
IEEE
School
School of Business and Law
RAS ID
83453
Funders
National Key Research and Development Program of China (2022YFB2702100) / National Natural Science Foundation of China (62472311, 62225203, U21A20516) / Key Research and Development Program of Ningxia (2023BEG02067) / Australia Research Council
Grant Number
ARC Number : DP240101591
Copyright
subscription content
Comments
Guo, W., Li, Z., Wang, X., Chen, Z., Zhao, J., Li, J., & Yuan, Y. (2025). CONVD: Attention enhanced dynamic convolutional embeddings for knowledge graph completion. IEEE Transactions on Knowledge and Data Engineering, 37(9), 5049–5062 https://doi.org/10.1109/TKDE.2025.3582243