A neural knowledge graph evaluator: Combining structural and semantic evidence of knowledge graphs for predicting supportive knowledge in scientific QA

作者：

Highlights：

• Inspired by cognitive studies, a new method exploits structure and semantic evidence.

• A comprehensive view of structures in knowledge graphs which was under-explored.

• Propose features to measure local and global structural dynamics of knowledge graphs.

• Propose a graph neural network to encode the structure and semantics of knowledge.

• Performance gains achieved by the NKGE on both structural and semantic features.

摘要

•Inspired by cognitive studies, a new method exploits structure and semantic evidence.•A comprehensive view of structures in knowledge graphs which was under-explored.•Propose features to measure local and global structural dynamics of knowledge graphs.•Propose a graph neural network to encode the structure and semantics of knowledge.•Performance gains achieved by the NKGE on both structural and semantic features.

论文关键词：Graph neural networks,Knowledge graph,Network analysis,Scientific question answering,Text entailment analysis

论文评审过程：Received 6 December 2019, Revised 13 March 2020, Accepted 18 May 2020, Available online 28 May 2020, Version of Record 28 May 2020.

论文官网地址：https://doi.org/10.1016/j.ipm.2020.102309