现代电子技术2024，Vol.47Issue(14)：41-45,5.DOI:10.16652/j.issn.1004-373x.2024.14.007
融合实体语义的实体关系抽取联合解码
Joint decoding for entity relation extraction with integrated entity semantics
摘要
Abstract
In allusion to the problem of polysemous words or entities with weak contextual connections in complex contexts,which makes it difficult for the model to recognize their relationships correctly,an entity relationship extraction model based on BERT and joint decoding is proposed.In this model,the BERT(bidirectional encoder representations from transformers)is used to semantically encode entities and extract their contextual information.Then,the self attention mechanism is used to label the head entity and predict the tail entity.A joint decoding mechanism is designed to combine entity semantic information and relationship extraction tasks for joint decoding.The experimental results show that,in comparison with the benchmark model,the proposed model can improve the accuracy and F1 value on the New York times(NYT)dataset and WebNLG dataset,effectively improving the accuracy of entity relationship extraction.关键词
实体关系抽取/实体语义/BERT/联合编码/自注意力机制/知识图谱Key words
entity relation extraction/entity semantics/BERT/joint decoding/self attention mechanism/knowledge graph分类
电子信息工程引用本文复制引用
张鑫,张思佳..融合实体语义的实体关系抽取联合解码[J].现代电子技术,2024,47(14):41-45,5.基金项目
辽宁省教育厅高等学校基本科研项目面上项目(LJKNZ20221095) （LJKNZ20221095）
辽宁省教育科学"十四五"规划课题(JG21DB076) （JG21DB076）
