一种煤矿机电设备领域稀疏样本的三元组抽取方法及装置
By combining BERT and Bi-LSTM, the problems of data sparsity and triple overlap in the field of coal mine electromechanical equipment were solved, achieving efficient triple extraction and improving the model's recognition performance and generalization ability.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIJING INFORMATION SCI & TECH UNIV
- Filing Date
- 2025-02-28
- Publication Date
- 2026-07-17
AI Technical Summary
Data sparsity and triple overlap issues in the field of coal mine electromechanical equipment have resulted in poor training and extraction performance of existing models, making it difficult to meet the needs of efficient management.
The BERT model is used for text vectorization, combined with Bi-LSTM to fuse Span contextual semantic relations, and a relation classifier is used for triple extraction. A sliding window mechanism is used to process entity representations, and a dataset and dictionary for the field of coal mine electromechanical equipment are constructed for data augmentation.
It improves the accuracy of triple extraction and the generalization ability of the model, especially when dealing with complex entities and overlapping triples, with an F1 value of 67.41%, which significantly improves the triple extraction effect in the field of coal mine electromechanical equipment.
Smart Images
Figure CN120508662B_ABST