基于调整方向感知的中文命名实体识别

C-ADA:Chinese named entity recognition based on adjusted direction-aware

在Transformer-CRF的基础上,通过调整函数优化方向感知和距离感知,提出一种基于调整函数动态调整方向感知的中文命名实体识别(C-ADA)模型.将C-ADA模型与BiLSTM、IDCNN、Transformer、TENER 4类模型在F_(1)得分上进行对比,结果表明,C-ADA模型均优于其他模型,在一定程度上验证了C-ADA具有较好的识别性能.

Based on Transformer-CRF and through optimizing,direction perception and distance perception by adjustment function,a Chinese named entity recognition based on adjusted direction-aware(C-ADA)model is proposed.In order to verify the effectiveness of the proposed method,C-ADA model is compared with the BiLSTM,IDCNN,Transformer,TENER,in terms of F_(1) score.The C-ADA model is proved to be superior to other models and has better recognition performance.