Relation extraction for manufacturing knowledge graphs based on feature fusion of attention mechanism and graph convolution network

作者：

Highlights：

• MKREM model is designed for relation extraction of manufacturing knowledge graphs.

• The initialization module is designed to initialize and quick mine hidden information.

• MSG and relation modules are proposed to mine dependency and semantic features, respectively.

• Feature fusion strategies are adopted to further improve the performance of MKREM.

• The F1 values of C-MKREM are 80.5 and 79.1 respectively, which is higher than the existing models.

摘要

•MKREM model is designed for relation extraction of manufacturing knowledge graphs.•The initialization module is designed to initialize and quick mine hidden information.•MSG and relation modules are proposed to mine dependency and semantic features, respectively.•Feature fusion strategies are adopted to further improve the performance of MKREM.•The F1 values of C-MKREM are 80.5 and 79.1 respectively, which is higher than the existing models.

论文关键词：Knowledge graph,Relation extraction,Graph convolution network (GCN),Attention mechanism,Manufacturing

论文评审过程：Received 11 March 2022, Revised 11 August 2022, Accepted 12 August 2022, Available online 18 August 2022, Version of Record 1 September 2022.

论文官网地址：https://doi.org/10.1016/j.knosys.2022.109703