Learning knowledge graph embeddings by deep relational roto-reflection

作者：

Highlights：

• Neural link prediction methods usually utilize the regular planner convolution.

• A neural model is proposed to use transformations for encoding relational types.

• The results show that deep transformations lead to performance improvement.

摘要

•Neural link prediction methods usually utilize the regular planner convolution.•A neural model is proposed to use transformations for encoding relational types.•The results show that deep transformations lead to performance improvement.

论文关键词：Knowledge graph,Link prediction,Convolutional neural network,Group convolution,Rotations,Reflections

论文评审过程：Received 16 March 2022, Revised 29 June 2022, Accepted 8 July 2022, Available online 21 July 2022, Version of Record 1 August 2022.

论文官网地址：https://doi.org/10.1016/j.knosys.2022.109451