Research on Named Entity Recognition in Operation System Based on Data Augmentation
WANG Xiaolong
WANG Nuanchen
MU Ge
YAN Xiaopei
LI Xinjin
Abstract：Named entity recognition(NER)for warfare systems encounters challenges such as difficulties in complex text structures,ambiguous entity boundaries,and imbalanced entity counts.To address these issues,a data augmentation-based NER method for warfare systems is proposed.Firstly,a corpus of warfare systems is constructed based on open-source internet data.The entity types of warfare systems are determined according to the United States department of defense architecture framework(DODAF)meta-model,and annotation rules are standardized in combination with corpus characteristics.Secondly,a deep learn-ing model based on RoBERTa-BILSTM-CRF is constructed for entity recognition within warfare systems.Finally,a data augmenta-tion approach leveraging large language model(LLM)is designed to enhance the original data from both grammatical and semantic perspectives,addressing the issue of imbalanced entity quantities in the warfare system corpus and improving the model's recogni-tion performance.Experimental results demonstrate that the model trained on the augmented dataset exhibits improvements in accu-racy,recall and F1 score.
Keywords：named entity recognitionLLMdata augmentationdeep learning
Publication Date：2025-05-20
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：6( 23-27,106 )