Abstract: To address data security risks arising from the rapid expansion of unstructured text data in railway passenger transport, this paper designed and developed an unstructured data governance and sensitive information recognition system for railway passenger tickets. This study leveraged full-traffic audit services to scan and identify multi-source file systems, developed an access engine adapted to heterogeneous file systems, and implemented metadata collection and content parsing of unstructured text data through standard access interfaces. The study built a rule base in accordance with sensitive information specifications and integrated regular expression matching, keyword recognition and the Bidirectional Encoder Representations from Transformers (BERT) model to achieve sensitive information recognition under complex semantic scenarios. The system has been deployed on China Railway Ticketing and Reservation System 7.0. It inventories unstructured text data assets, achieves an automatic sensitive information recognition accuracy of over 92.3%, efficiently identifies sensitive information within unstructured text, and provides technical support for data access permission control.