Negative sampling and rule mining for explainable link prediction in knowledge graphs
作者：
Highlights：
• We propose a simple but efficient negative triple sampling method for KG called SNS.
• We designed SNS as a general method that can be plugged to any KG embedding model.
• We propose an efficient rule mining method on a KG and its embedding.
• We extensively evaluate our rule mining method in combination with the SNS method.
• We provide a strategy to use mined rules to explain embedding-based link predictions.
摘要
•We propose a simple but efficient negative triple sampling method for KG called SNS.•We designed SNS as a general method that can be plugged to any KG embedding model.•We propose an efficient rule mining method on a KG and its embedding.•We extensively evaluate our rule mining method in combination with the SNS method.•We provide a strategy to use mined rules to explain embedding-based link predictions.
论文关键词：Knowledge graph embedding,Link prediction,Negative sampling,Rule mining,Explainability
论文评审过程：Received 20 February 2022, Revised 14 May 2022, Accepted 16 May 2022, Available online 24 May 2022, Version of Record 1 June 2022.
论文官网地址：https://doi.org/10.1016/j.knosys.2022.109083
