Enhanced Named Entity Recognition for Drilling Engineering Using Masked CRF

Drilling engineering reports are essential for oil and gas extraction, documenting critical geological information and operational parameters. However, their unstructured format, diverse writing styles, and long-distance dependencies between related entities pose challenges for automated information extraction. Traditional rule-based and machine learning methods struggle with domain-specific terminology, inconsistent entity structures, and complex contextual relationships, leading to low accuracy and poor generalization. To address these challenges, we introduce Masked Conditional Random Fields (Masked CRF) to prevent inverted sequence errors and enhance contextual understanding and sequence labeling consistency. To overcome the lack of domain-specific annotated data, we establish an annotation paradigm tailored for drilling engineering reports so that obtains a specialized drilling corpus with 12 entity categories and 20,727 labeled instances. Experimental results demonstrate that our approach achieves an F1-score of 84.09%, outperforming the base model by 2.05, particularly improving the recognition of long-tail and complex entities. The novel approach improves NER for drilling engineering, providing engineers with an efficient tool for extracting key information, accelerating data analysis, and improving decision-making in drilling operations.