Multi-granularity sequential neural network for document-level biomedical relation extraction
作者：
Highlights：
• We propose a sequential network via multi-granularity information to extract document-level biomedical entity relation, which tackles long-distance dependencies and complex contexts causing by numerous alias and inter-sentence relation.
• The proposed method can focus on entity pairs that precisely reflect the biomedical entity relation.
• Multi-granularity structural information is shown to be effective for document-level biomedical relation extraction by some analytical experiments.
摘要
•We propose a sequential network via multi-granularity information to extract document-level biomedical entity relation, which tackles long-distance dependencies and complex contexts causing by numerous alias and inter-sentence relation.•The proposed method can focus on entity pairs that precisely reflect the biomedical entity relation.•Multi-granularity structural information is shown to be effective for document-level biomedical relation extraction by some analytical experiments.
论文关键词：Multi-granularity information,Document-level biomedical relation extraction,Sequential neural network
论文评审过程：Received 26 April 2021, Revised 5 August 2021, Accepted 7 August 2021, Available online 18 August 2021, Version of Record 18 August 2021.
论文官网地址：https://doi.org/10.1016/j.ipm.2021.102718
