Extraction of production process standard entity relationships based on MacBERT-BiGRU-IDCNN-CRF
Yiming Xu · Procedia Computer Science · 2025
Production process standard texts are characterized by high knowledge density, often containing a large number of nested entities, ambiguous boundaries, and overlapping relationships in process specifications and requirements. To address these issues, the paper proposes a deep learning-based method for entity and relation extraction from production process texts, constructing a MacBERT-BiGRU-IDCNN-CRF combined neural network model for extracting information from production process standard texts. The model first utilizes the MacBERT pre-trained language model to vectorize production process knowledge, extracting character-level and sentence-level features. Then, the BiGRU and IDCNN neural networks are employed to fuse character and sentence features, enhancing the model’s semantic understanding capabilities. Finally, the CRF model outputs the globally optimal label sequence. To validate the effectiveness of the proposed method, experiments were conducted on a production process standard entity-relation dataset. The results demonstrate that the model achieves a precision of 0.93, a recall of 0.961, and an F1 score of 0.945, outperforming traditional entity recognition models. The research presented in this paper provides a valuable reference for the construction of domain-specific knowledge graphs.