To address the data security risks arising from the explosive growth of railway passenger transport data, the core lies in achieving intelligent identification and dynamic protection of sensitive information. Then, an intelligent identification technology for sensitive data in railway passenger tickets based on data knowledge base is proposed. Firstly, a three-level knowledge base of "laws and regulations-industry standards-enterprise norms" is constructed. Secondly, combined with historical railway passenger ticket data, a multi-level intelligent identification algorithm for sensitive data is designed, thereby efficiently and accurately identifying sensitive information in multi-modal data. On this basis, the graph technology is finally introduced to construct a data asset and sensitive data lineage graph, and based on the topological relationship of data flow, the efficient propagation of sensitive information labels among related data nodes is achieved. The results show that the sensitive information identification efficiency of the proposed technology reaches about 217 000 messages per second in structured data processing, which is almost twice as high as the traditional solution. In unstructured data processing, through domain knowledge graphs injection, the F1 value of sensitive entity recognition is increased to 91.24%, and the context misjudgment rate is reduced to 5.88%. The accuracy of text extraction and sensitive information recognition of multimedia images reaches 93.71%. This technology can significantly improve the accuracy and processing efficiency of sensitive data identification in railway passenger tickets.
图像中敏感文字信息检测的研究方法主要可分为2类:基于传统视觉特征的敏感信息检测和基于图像文本特征的敏感信息检测。Krasser等[7]重点考虑图像的边缘方向一致性矢量、尺度不变特征变换特征以及颜色直方图,将这些视觉特征作为线性支持向量机分类器的判别标准。Wang等[8]将线分布云与最大稳定极值区域算法相结合以提取文本区域,该方法在弱光条件下仍具有较好的检测效果。随着光学字符识别(Optical Character Recognition,OCR)技术的不断发展,CRNN算法[9]和基于注意力机制的文本识别算法已成为当前主流。CRNN算法先通过卷积神经网络提取图像的空间特征,再利用循环神经网络将空间特征转化为序列特征,最终输出对应的文本内容,从而提升识别性能;基于注意力机制的识别算法通过引入注意力权重矩阵,根据输入序列的特征动态计算每个元素的重要性,实现更精准的文字识别。
SHANXinghua, ZHANGZhiqiang, NINGFei, et al. Key Technology Application and System Implementation of China Railway Electronic Ticket [J]. China Railway Science, 2021, 42 (5): 162-173. in Chinese
FUZefan, YAOJingfa, TENGGuifa. Research on Website Sensitive Information Identification and Variant Restoration Technology Based on BERT Model [J]. Modern Electronics Technique, 2024, 47 (23): 105-112. in Chinese
LIYang, PANQuan, YANGTao. Sensitive Information Recognition Based on Short Text Sentiment Analysis [J]. Journal of Xi’an Jiaotong University, 2016, 50 (9): 80-84. in Chinese
[7]
LIW P, WUH Y, YANGJ. Intelligent Recognition Algorithm for Social Network Sensitive Information Based on Classification Technology [J]. Discrete and Continuous Dynamical Systems-S, 2019, 12 (4/5): 1385-1398.
[8]
XUG, WUX, YAOH, et al. Research on Topic Recognition of Network Sensitive Information Based on SW-LDA Model [J]. IEEE Access, 2019, 7: 21527-21538.
[9]
XUG, YUZ, CHENZ, et al. Sensitive Information Topics-Based Sentiment Analysis Method for Big Data [J]. IEEE Access, 2019, 7: 96177-96190.
[10]
KRASSERS, TANGY, GOULDJ, et al. Identifying Image Spam Based on Header and File Properties Using C4.5 Decision Trees and Support Vector Machine Learning [C]// 2007 IEEE SMC Information Assurance and Security Workshop. New York: IEEE, 2007: 255-261.
[11]
WANGW, WUY, PALAIAHNAKOTES, et al. Cloud of Line Distribution for Arbitrary Text Detection in Scene/Video/License Plate Images [C]// Advances in Multimedia Information Processing-PCM 2017. Cham: Springer International Publishing, 2018: 433-443.
[12]
SHIB, BAIX, YAOC. An End-to-End Trainable Neural Network for Image-Based Sequence Recognition and Its Application to Scene Text Recognition [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39 (11): 2298-2304
WANGQing, CHENJie. A Review of Regular Expression Matching Research in Deep Packet Inspection Technology [J]. Network Security Technology & Application, 2024 (5): 28-30. in Chinese
JIANGHaiyang, LIXuefei, YANGYe, et al. Distance Comparison Based Parallel Pattern Matching [J]. Journal of Electronics & Information Technology, 2022, 44 (2): 581-590. in Chinese
YANGHong, MENGXiaokai, YUHua, et al. Research on Primary Equipment Defect Diagnosis Method Based on the BERT Model [J]. Power System Protection and Control, 2025, 53 (7): 155-164. in Chinese
BAIZhicheng, LIQing, CHENPeng, et al. Text Detection in Natural Scenes: a Literature Review [J]. Chinese Journal of Engineering, 2020, 42 (11): 1433-1448. in Chinese
RADFORDA, KIMJ W, HALLACYC, et al. Learning Transferable Visual Models from Natural Language Supervision [J]. ArXiv e-Prints, 2021: arXiv:2103.00020 [cs. CV].
[23]
ISLAMN S, RAHMANM W, JOSEJ, et al. High Performance RDMA-Based Design of HDFS over InfiniBand [C]// SC’12: Proceedings of the International Conference on High Performance Computing, Networking, Storage and Analysis. New York: IEEE, 2012.
[24]
MONDALS, MUKHERJEEN. Efficient NoSQL Graph Database for Storage and Access of Health Data [C]// Computer Communication, Networking and IoT. Singapore: Springer, 2021: 135-146.
PANXiaohua, JINYong, GAOYanghua, et al. Data Lineage Construction Method for Complex Data Audit Requirements [J]. Application Research of Computers, 2024, 41 (1): 76-82. in Chinese
[27]
XIEY R. Application of CRNN and OpenGL in Intelligent Landscape Design Systems Utilizing Internet of Things, Explainable Artificial Intelligence, and Drone Technology [J]. IEEE Transactions on Consumer Electronics, 2025, 71 (2): 3930-3940.