An efficiency relation-specific graph transformation network for knowledge graph representation learning

作者：

Highlights：

• We design a relation-specific graph transformation network (RGTN) for KGRL.

• The RGTN is able to transform the relation-based graph to a new path-based graph.

• The RGTN can aggregate multi-hop neighbors on the generated path-based graph.

• Experimental results show that the proposed RGTN achieves promising results.

摘要

•We design a relation-specific graph transformation network (RGTN) for KGRL.•The RGTN is able to transform the relation-based graph to a new path-based graph.•The RGTN can aggregate multi-hop neighbors on the generated path-based graph.•Experimental results show that the proposed RGTN achieves promising results.

论文关键词：Knowledge graph,Graph transformation,Representation learning,Graph neural networks

论文评审过程：Received 13 April 2022, Revised 24 July 2022, Accepted 28 August 2022, Available online 15 September 2022, Version of Record 15 September 2022.

论文官网地址：https://doi.org/10.1016/j.ipm.2022.103076