Research on Railway Standard Rule Identification and Dataset Generation Method Based on Natural Language Processing

Railway information systems have made significant progress in data standardization and normalization when addressing data quality issues. However, verifying whether business data complies with standard document requirements still relies heavily on manual parsing and structuring, which not only consumes considerable human resources but also risks operational errors due to the large volume of data being processed manually. To address this, this paper proposes an automated method based on OCR recognition and natural language processing, integrated with semantic analysis techniques. This approach accurately identifies the data rules and standard datasets within standard documents, converting them into executable rule expressions and standardized datasets. The method not only efficiently verifies the data quality covered by standard documents but also enables the batch generation of a unified data rule library and standard datasets for the railway industry. This significantly enhances the efficiency of data quality inspections and provides critical support for comprehensively improving railway data quality.