PathReasoner: Explainable reasoning paths for commonsense question answering
作者：
Highlights：
• A reasonable and explainable framework is proposed to explicitly incorporate external reasoning paths with structured information to explain and facilitate commonsense QA.
• We use a Hierarchical Path Learner including an intra-path encoder and an inter-path encoder to reason the path over questions.
• The experiments on two typical commonsense QA datasets demonstrate the effectiveness of our proposed PathReasoner.
摘要
•A reasonable and explainable framework is proposed to explicitly incorporate external reasoning paths with structured information to explain and facilitate commonsense QA.•We use a Hierarchical Path Learner including an intra-path encoder and an inter-path encoder to reason the path over questions.•The experiments on two typical commonsense QA datasets demonstrate the effectiveness of our proposed PathReasoner.
论文关键词：Commonsense reasoning,Knowledge graph,Reasoning path,Interpretability
论文评审过程：Received 16 May 2021, Revised 13 October 2021, Accepted 16 October 2021, Available online 29 October 2021, Version of Record 29 October 2021.
论文官网地址：https://doi.org/10.1016/j.knosys.2021.107612
