BERT-Based Joint Task Approach for Named Entity Recognition

Named Entity Recognition constitutes a fundamental task in natural language processing. To address the challenges of ambiguous entity boundaries and inaccurate type classification in Chinese NER, this paper proposes a BERT-based joint task model. The framework decouples entity detection and classification into distinct subtasks, while incorporating a gated-enhanced local attention focusing mechanism to prioritize potential entity spans. Experiments on the CLUENER fine-grained NER dataset and the Resume dataset demonstrate that the proposed method achieves superior F1 scores compared to existing models, validating its effectiveness.