Research on NER technology for railway employees’ work experience based on the BERT-BiLSTM-CRF model

With the rapid development of the railway industry, human resource management systems have accumulated massive textual data of employees’ work experience. To meet practical business requirements including talent evaluation, job assignment, and employee development planning, this study proposes a BERT-BiLSTM-CRF model for accurate named entity recognition (NER) of railway employees’ work experience texts. A multi-type entity annotation scheme for employee work experience is established. Domain knowledge enhancement and an early stopping training strategy are adopted to improve the model’s adaptability to railway professional texts. Experimental results demonstrate that the proposed model achieves excellent performance in the NER task of railway human resource work experience texts, with an F1-score of 98%. It significantly outperforms traditional machine learning methods and mainstream deep learning benchmark models.