江苏大学学报（自然科学版）2025，Vol.46Issue(4)：431-437,7.DOI:10.3969/j.issn.1671-7775.2025.04.009
基于多路局部特征整合的嵌套命名实体识别方法
Nested named entity recognition method based on multiplexed local feature integration
摘要
Abstract
To solve the problems of boundary ambiguity and difficult nested entity extraction in nested named entity recognition(NER),the nested named entity recognition method based on multiplexed local feature integration was proposed.By the proposed method,the sequential features were decomposed into forward and backward representations by the bidirectional long short-term memory network.The nested NER task was segmented by the entity length,and the fixed-length local features were integrated by the multi-scale convolutional networks.The forward and backward features were matched to generate prediction results.The pre-weighting mechanism was introduced to mitigate the inter-layer information transmission errors in deep architectures.The experiments were conducted on the ACE2005 and GENIA datasets for comparing the proposed method with the existing nested NER approaches.The results show that compared with the best Dependency Parsing method of other methods,the proposed model achieves superior performance with improved F1 scores by 0.18％and 0.03％on ACE2005 and GENIA,respectively.By the proposed method,a certain performance improvement can be achieved compared with the current state-of-the-art models.关键词
自然语言处理/嵌套命名实体识别/深度学习/卷积神经网络/长短时记忆网络/特征融合/自适应学习Key words
natural language processing/nested named entity recognition/deep learning/convolutional neural networks/long short-term memory network/feature fusing/adaptive learning分类
信息技术与安全科学引用本文复制引用
王进,蒋诗琪..基于多路局部特征整合的嵌套命名实体识别方法[J].江苏大学学报（自然科学版）,2025,46(4):431-437,7.基金项目
国家自然科学基金资助项目(62302074) （62302074）
