Learning attention-based representations from multiple patterns for relation prediction in knowledge graphs

作者：

Highlights：

• Attention-based Embeddings from Multiple Patterns of a Knowledge Graph to learn contextualized representations.

• Attention-enhanced message-passing to capture local semantics but focusing on different aspects of the entity neighborhood.

• Experimental results with large and small knowledge graphs.

摘要

•Attention-based Embeddings from Multiple Patterns of a Knowledge Graph to learn contextualized representations.•Attention-enhanced message-passing to capture local semantics but focusing on different aspects of the entity neighborhood.•Experimental results with large and small knowledge graphs.

论文关键词：Knowledge graphs,Representation learning,Embeddings,Attention mechanism

论文评审过程：Received 7 January 2022, Revised 6 June 2022, Accepted 7 June 2022, Available online 16 June 2022, Version of Record 2 July 2022.

论文官网地址：https://doi.org/10.1016/j.knosys.2022.109232