计算机科学 ›› 2026, Vol. 53 ›› Issue (6A): 250500048-6.doi: 10.11896/jsjkx.250500048
郑明坤, 庞春江, 王新颖
ZHENG Mingkun, PANG Chunjiang, WANG Xinying
摘要： 针对高压断路器领域中文实体识别任务中专业术语复杂、实体边界模糊及长文本依赖等问题,提出了一种基于 SoftLexicon-BERT-GlobalPointer 的混合模型方法。该方法利用 SoftLexicon 结合领域词典增强词表示能力,通过 BERT 预训练模型捕捉上下文语义,采用 GlobalPointer 解码策略优化实体边界预测。该方法在自主构建的高压断路器数据集上表现优异,F1 值达到 90.62%,显著优于传统 BiLSTM-CRF 及基础 BERT 模型,为电力设备智能运维提供了高效的 NLP 解决方案。
中图分类号:
| [1] LI J,SUN A,HAN J,et al.A Survey on Deep Learning for Named Entity Recognition:Extended Abstract[C]//2023 IEEE 39th International Conference on Data Engineering(ICDE).Anaheim,CA,USA,2023:3817-3818. [2] QUAN Y Q,ZHANG H T,YANG B,et al.A Survey of Named Entity Recognition Based on Deep Learning[J].Microelectronics & Computer,2026,43(2):8-21. [3] SUN Z Y,CHEN L,WEI S B,et al.Chinese Named Entity Recognition Based on Boundary Information and Lexical Information Enhancement[J].Journal of Nanjing Normal University (Engineering and Technology Edition),2024,24(4):79-86. [4] ZHANG T.Research on Chinese Medical Named Entity Recognition Based on MedBERT-wwm and GlobalPointer[D].Beijing:Beijing University of Chemical Technology,2024. [5] LUO L,XIA Q X.Chinese NER Model Fusing ERNIE and Improved Transformer[J].Computer Technology and Development,2022,32(10):120-125. [6] HAQUE M Z,ZAMAN S,SAURAV J R,et al.B-NER:A Novel Bangla Named Entity Recognition Dataset With Largest Entities and Its Baseline Evaluation[C]//IEEE Access.2023:45194-45205. [7] YUAN M,LI Y.Bi-Lattice LSTM Model with Self-Attentionfor Chinese NER[C]//2020 IEEE 20th International Confe-rence on Communication Technology(ICCT).2020:1465-1470. [8] MA R T,PENG M L,ZHANG Q,et al.Simplify the Usage of Lexicon in Chinese NER[C]//Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics.2020:5951-5960. [9] GAO Y,LI H,ZHONG W H.Chemical Safety Accident Named Entity Recognition Based on Self-attention and SoftLexicon[J].Journal of Computer Applications,2024,44(S2):61-65. [10] CHEN C,SHI H,XU Y,et al.Text Classification of Oil Field Safety Production Hazards Based on BERT-BiLSTM[J].Science Technology and Engineering,2024,24(29):12650-12657. [11] ZHOU L Q,LI Y,HUA Y P,et al.Entity-Relation Extraction of Oil and Gas Knowledge Graph Based on BERT-Graph-GlobalPointer[J].Electronic Components and Information Technology,2024,8(9):63-68. [12] ZHENG X Y,CHEN T,ZHOU J J.Geographic Named Entity Extraction Model for Surface Water System Using Extended Word Embedding BERT[J].Geospatial Information,2025,23(2):1-6. [13] WANG Y,ZHOU Y,QI H,et al.A complex history browsing text categorization method with improved BERT embedding la-yer[J/OL].https://doi.org/10.1007/s10489-025-06298. [14] BORTHAKUR S,SARVAIYA J N.NER for Extracting Tourist Attractions from Textual Information:A BERT-BiLSTM-CRF Approach[C]//2023 IEEE Fifth International Conference on Advances in Electronics,Computers and Communications(ICAECC).Bengaluru,India,2023:1-5. [15] ZHANG J Y.Research on Chinese Electronic Medical Record Named Entity Recognition Based on GlobalPointer and Multi-feature Fusion[D].Wuhan:Huazhong University of Science and Technology,2023. [16] LI M J,LI W J,WANG H R.Entity Recognition Fusing Lexical Information and GlobalPointer[J].Journal of Shandong University (Engineering Science),2024,54(1):91-99. [17] MA Y.Analysis of Software Development System Based on Natural Language Processing[J].Network Security Technology & Application,2025(3):69-72. [18] WANG H,ZHANG C,ZHANG B,et al.SoftLexicon-BERT-BiLSTM-CRF-based Named Entity Recognition Model for Transportation Organization Design Domain[C]//2023 International Conference on Image Processing,Computer Vision and Machine Learning(ICICML).2023:203-207. |
|
||
首页