Abstract
With the acceleration of digitalization, financial data of listed companies are facing complex and diverse security risks and privacy leakage hazards. To improve the accuracy and effectiveness of data security management, this study proposes a financial data security risk prediction and privacy protection method for listed companies based on hypergraph learning. It is based on hypergraph learning to construct a risk prediction model, which optimizes the hyperedge weight algorithm to explore the complex relationships between multi-dimensional financial data, and integrates bidirectional gated loop units to achieve dynamic risk prediction. A layered privacy protection strategy based on differential privacy has been designed, which adaptively allocates privacy budget according to data sensitivity. The experiment showed that this risk prediction method had an accuracy rate of 93.1% on financial datasets, which was 9.2% higher than traditional graph attention networks. The false positive rate was only 2.8%, and it could accurately identify 8 potential security risks. When the privacy budget was 0.6, the privacy protection method controlled the data utility loss at 11.5%, which improved the information utilization rate by 25.3% compared to traditional generalized anonymity methods. It also met the privacy compliance standards of the financial industry in six scenarios, including data transactions and internal audits. Through the coordinated design of risk prediction and privacy protection, this study establishes a comprehensive financial data security management framework for listed companies. The method’s core innovation lies in systematically resolving the balance between security safeguards and data utility. By integrating hypergraph learning with BiGRU prediction models, it achieves high-precision dynamic risk alerts for complex scenarios while preserving data analytical value. Additionally, the adaptive hierarchical protection strategy based on differential privacy provides compliant and measurable security for sensitive information. This system delivers a practical, integrated technical solution for the financial industry that balances precise risk alerts with privacy controllability.
References
Wang J, Zhou L, Jiang C, Wang Z. Modeling and Interpreting the Propagation Influence of Neighbor Information in Time-Variant Networks with Exemplification by Financial Risk Prediction. Journal of Management Information Systems, 2025, 42(1): 105–142. DOI: 10.1080/07421222.2025.2452016.
Yuan L, Duan M, Xiao G, Tang Z, Li K. BM-FL: A Balanced Weight Strategy for Multi-Stage Federated Learning Against Multi-Client Data Skewing. IEEE Transactions on Knowledge and Data Engineering, 2024, 36(9): 4486–4498. DOI: 10.1109/TKDE.2024.3372708.
He Y, Tai W, Zhou F, Yang Y. Exploring Hypergraph of Earnings Call for Risk Prediction (Student Abstract). Proceedings of the AAAI Conference on Artificial Intelligence, 2024, 37(13): 16226–16227. DOI: 10.1609/aaai.v37i13.26973.
Wang S, Dong Y, Zhang B, Chen Z, Fu X, He Y, Shen C, Zhang C, Chawla N V, Li J. Safety in Graph Machine Learning: Threats and Safeguards. ArXiv, 2024. DOI: 10.48550/arXiv.2405.11034.
Das S, Huang X, Adeshina S, Yang P, Bachega L. Credit Risk Modeling with Graph Machine Learning. INFORMS Journal on Data Science, 2023, 2(2): 197–217. DOI: 10.1287/ijds.2022.00018.
Wang Z, Jiang C, Zhao H. Depicting Risk Profile Over Time: A Novel Multiperiod Loan Default Prediction Approach. MIS Quarterly, 2023, 47(4): 1455–1485. DOI: 10.25300/MISQ/2022/17491.
Effendi F, Chattopadhyay A. Privacy-Preserving Graph-Based Machine Learning with Fully Homomorphic Encryption for Collaborative Anti-Money Laundering. International Conference on Security, Privacy, and Applied Cryptography Engineering, 2024, pp. 80–105. Springer Nature Switzerland, Cham. DOI: 10.1007/978-3-031-80408-3_6.
Jia J, Yang L, Wang Y, Sang A. Hyper Attack Graph: Constructing a Hypergraph for Cyber Threat Intelligence Analysis. Computers & Security, 2025, 149: 104194. DOI: 10.1016/j.cose.2024.104194.
Xu X. Empirical asset pricing based on network big data mining and privacy protection. Neural Computing and Applications, 2025, 37(12): 7841–7855. DOI: 10.1007/s00521024101108.
Matalobos Veiga J M, Criado R, Romance Del Rio M, Iglesias Perez S, Partida Rodriguez A, Hanumanthappa Manjunatha K K. A Hypergraph-based Model for Cyberincident-related Data Analysis. In Proceedings of the 2024 European Interdisciplinary Cybersecurity Conference, 2024: 161–162. DOI: 10.1145/3655693.3661300.
Mohapatra D, Bhoi S K, Jena K K, Sahoo K S, Nayyar A, Shah M A. Rank-Label Anonymization for the Privacy-Preserving Publication of a Hypergraph Structure. IEEE Access, 2022, 10: 118253–118267. DOI: 10.1109/ACCESS.2022.3219107.
Li Z, Wang C, Wang X, Chen Z, Li J. HJE: Joint Convolutional Representation Learning for Knowledge Hypergraph Completion. IEEE Transactions on Knowledge and Data Engineering, 2024, 36(8): 3879–3892. DOI: 10.1109/TKDE.2024.3365727.
Yang Y, Qin Y, Fan Y, Zhang Z. Unlocking the Power of Voice for Financial Risk Prediction: A Theory-Driven Deep Learning Design Approach. MIS Quarterly, 2023, 47(1): 63–96. DOI: 10.25300/MISQ/2022/17062.
Huang J, Lei F, Jiang J, Zeng X, Ma R, Dai Q. Multi-Order Hypergraph Convolutional Networks Integrated with Self-Supervised Learning. Complex & Intelligent Systems, 2023, 9(4): 4389–4401. DOI: 10.1007/s40747-022-00964-7.
Su L, Gan Z, Wang J, Li X. Structural Complementary Hypergraph Defense Framework against Adversarial Attacks. Knowledge-Based Systems, 2025, 114897. DOI: 10.1016/j.knosys.2025.114897.
Yuan H, Xu J, Huang R, Song M, Wang C, Yang Y. Can Graph Neural Networks Expose Training Data Properties? An Efficient Risk Assessment Approach. Advances in Neural Information Processing Systems, 2024, 37: 69361–69385. DOI: 10.52202/079017-2215.
Tai B, Tsou Y, Li S, Huang Y, Tsai P, Tsai Y. User-Driven Synthetic Dataset Generation With Quantifiable Differential Privacy. IEEE Transactions on Services Computing, 2023, 16(5): 3812–3826. DOI: 10.1109/TSC.2023.3287239.
Liu J, Ge Z, Wang Y. Role of Environmental, Social, and Governance Rating Data in Predicting Financial Risk and Risk Management. Corporate Social Responsibility and Environmental Management, 2024, 31(1): 260–273. DOI: 10.1002/csr.2567.
Zhang P, Sun H, Zhang Z, Cheng X, Zhu Y, Zhang J. Privacy-Preserving Recommendations With Mixture Model-Based Matrix Factorization Under Local Differential Privacy. IEEE Transactions on Industrial Informatics, 2025, 21(7): 5451–5459. DOI: 10.1109/TII.2025.3555993.
Markovic T, Leon M, Buffoni D, Punnekkat S. Random Forest with Differential Privacy in Federated Learning Framework for Network Attack Detection and Classification. Applied Intelligence, 2024, 54(17–18): 8132–8153. DOI: 10.1007/s10489-024-05589-6.
Elhoseny M, Metawa N, Sztano G, Elhasnony I M. Deep Learning-Based Model for Financial Distress Prediction. Annals of Operations Research, 2025, 345(2–3): 885–907. DOI: 10.1007/s10479-022-04766-5.
Li Z, Chen Y, Wang X, Yao L, Xu G. Multi-view GCN for Loan Default Risk Prediction. Neural Computing and Applications, 2024, 36(20): 12149–12162. DOI: 10.1007/s00521-024-09695-x.
Chen C, Zhang X, Lin S, Chen X. MPGStack: Membership Privacy Protection on Graph Data via Model Stacking. IEEE Transactions on Dependable and Secure Computing, 2025, 22(4): 3723–3736. DOI: 10.1109/TDSC.2025.3541403.
Asif H, Vaidya J, Papakonstantinou P A. Identifying Anomalies While Preserving Privacy. IEEE Transactions on Knowledge and Data Engineering, 2023, 35(12): 12264–12281. DOI: 10.1109/TKDE.2021.3129633.
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
Copyright (c) 2026 Journal of Cyber Security and Mobility