Automated Compliance Review of Data Assets a Program Combining Natural Language Processing Techniques and Improved YOLO
摘要
By analyzing the problems of automated Data Asset compliance review, an automated compliance review scheme combining improved YOLO algorithm and Natural Language Processing technology is proposed. The method combines rule-based exact matching, semantic similarity calculation based on fine-tuned BERT model, and image text recognition based on YOLOv10, and constructs a rule-semantic-visual tri-modal fusion architecture. The results show that the system can improve the average auditing accuracy and auditing time efficiency significantly compared with manual auditing, and it can handle documents in multiple formats with good robustness. This study provides a feasible solution for automated compliance review of Data Assets, which significantly improves review efficiency, flexibility, and accuracy.