MHGR-Net: A Multimodal Heterogeneous Graph Transformer Approach for Compliance Risk Identification and Early Warning in Business-Finance-Law-Tax Domains
Abstract
To address the challenges associated with highly heterogeneous Business–Finance–Law–Tax (BFLT) data and increasingly complex risk propagation patterns, this study proposed a Multimodal and Heterogeneous Graph-based Risk Network (MHGR-Net). The framework consisted of three stages. First, a Universal Information Extraction (UIE) model integrated document layout information with semantic features. Second, a temporal heterogeneous knowledge graph was constructed to model dynamic risk associations. Third, a Heterogeneous Graph Transformer (HGT) was employed to learn risk path weights through a self-attention mechanism. The experimental dataset included more than 13,000 multimodal documents, such as contracts, financial statements, and judicial judgments, collected from 30 Chinese A-share listed companies. MHGR-Net achieved an Area Under the Curve (AUC) of 94.1% and an accuracy of 95.03%. The model required only 14 minutes for training. It outperformed mainstream baseline methods, including XGBoost, Graph Convolutional Network (GCN), and Temporal Graph Convolutional Network (T-GCN), in both predictive performance and computational efficiency. The proposed system reduced data silos and provided reliable decision support for enterprises shifting from passive compliance management to proactive and intelligent risk governance.References
“The urgency of reforming Indonesia’s tax Law in the face of economic digitalization,” Cogent Social Sciences, vol. 9, no. 2, pp. 2285242, 2023.
[2] N. Xu, W. Lv and J. Wang, “The impact of digital transformation on firm performance: a perspective from enterprise risk management,” Eurasian Business Review, vol. 14, no. 2, pp. 369–400, 2024.
[3] K. Yaffe, E. Vittinghoff, S. Dublin, C. B. Peltz, L. F. Fleckenstein, D. E. Rosenberg, et al., “Effect of personalized risk-reduction strategies on cognition and dementia risk profile among older adults: the SMARRT randomized clinical trial,” JAMA Internal Medicine, vol. 184, no. 1, pp. 54–62, 2024.
[4] B. Solikhah, C. L. Chen, P. Y. Weng and M. A. S. Al-Faryan, “Related party transactions and tax avoidance: does government ownership play a role?” Corporate Governance: The International Journal of Business in Society, vol. 25, no. 4, pp. 763–785, 2025.
[5] W. Gleißner and T. B. Berger, “Enterprise risk management: improving embedded risk management and risk governance,” Risks, vol. 12, no. 12, pp. 196, 2024.
[6] M. L. D. Tewu, I. Bernarto, S. Suwarno and P. Lisdiono, “The effect of enterprise risk management and compliance practices on the firm performance of Indonesian banking companies,” Indonesian Journal of Business and Entrepreneurship, vol. 10, no. 1, pp. 52, 2024.
[7] X. Li, J. Wang and C. Yang, “Risk prediction in financial management of listed companies based on optimized BP neural network under digital economy,” Neural Computing and Applications, vol. 35, no. 3, pp. 2045–2058, 2023.
[8] N. Tamascelli, G. E. Scarponi, M. T. Amin, Z. Sajid, N. Paltrinieri, F. Khan, et al., “A neural network approach to predict the time-to-failure of atmospheric tanks exposed to external fire,” Reliability Engineering & System Safety, vol. 245, pp. 109974, 2024.
[9] T. Wang, J. Guo, Y. Shan, Y. Zhang, B. Peng and Z. Wu, “A knowledge graph–GCN–community detection integrated model for large-scale stock price prediction,” Applied Soft Computing, vol. 145, pp. 110595, 2023.
[10] Y. Shi, Y. Wang, Y. Qu and Z. Chen, “Integrated GCN-LSTM stock prices movement prediction based on knowledge-incorporated graphs construction,” International Journal of Machine Learning and Cybernetics, vol. 15, no. 1, pp. 161–176, 2024.
[11] X. Wang and N. El-Gohary, “Deep learning-based relation extraction and knowledge graph-based representation of construction safety requirements,” Automation in Construction, vol. 147, pp. 104696, 2023.
[12] P. Yang, H. Wang, Y. Huang, S. Yang, Y. Zhang, L. Huang, et al., “LMKG: A large-scale and multi-source medical knowledge graph for intelligent medicine applications,” Knowledge-Based Systems, vol. 284, pp. 111323, 2024.
[13] M. Shim, H. Choi, H. Koo, K. Um, K. H. Lee and S. Lee, “OmEGa (Ω): Ontology-based information extraction framework for constructing task-centric knowledge graph from manufacturing documents with large language model,” Advanced Engineering Informatics, vol. 64, pp. 103001, 2025.
[14] H. Ding, H. Hou, L. Wang, X. Cui, W. Yu and D. I. Wilson, “Application of convolutional neural networks and recurrent neural networks in food safety,” Foods, vol. 14, no. 2, pp. 247, 2025.
[15] Z. Wang, “Application of CNN-based financial risk identification and management convolutional neural networks in financial risk,” Systems and Soft Computing, vol. 7, pp. 200215, 2025.
[16] K. F. Mojdehi, B. Amiri and A. Haddadi, “A novel hybrid model for credit risk assessment of supply chain finance based on topological data analysis and graph neural network,” IEEE Access, vol. 13, pp. 13101–13127, 2025.
[17] S. Chen, S. Liu and J. Liu, “Type-specific modality alignment for multi-modal information extraction,” IEEE Signal Processing Letters, vol. 31, pp. 1525–1529, 2024.
[18] X. Li, J. Jin, Y. Zhou, Y. Zhang, P. Zhang, Y. Zhu, et al., “From matching to generation: A survey on generative information retrieval,” ACM Transactions on Information Systems, vol. 43, no. 3, pp. 1–62, 2025.
[19] G. Cai, X. Zheng, J. Guo and W. Gao, “Real-time identification of borehole rescue environment situation in underground disaster areas based on multi-source heterogeneous data fusion,” Safety Science, vol. 181, pp. 106690, 2025.
[20] S. Yao, Q. Wu, Q. Kang, Y. W. Chen and Y. Lu, “An interpretable XGBoost‐based approach for Arctic navigation risk assessment,” Risk Analysis, vol. 44, no. 2, pp. 459–476, 2024.
[21] X. Li, X. Li, J. Jia, L. Li, J. Yuan and Y. Gao, “A high accuracy and adaptive anomaly detection model with dual-domain graph convolutional network for insider threat detection,” IEEE Transactions on Information Forensics and Security, vol. 18, pp. 1638–1652, 2023.
[22] X. Liu, J. Lu, X. Chen, Y. H. C. Fong, X. Ma and F. Zhang, “Attention based spatio-temporal graph convolutional network with focal loss for crash risk evaluation on urban road traffic network based on multi-source risks,” Accident Analysis & Prevention, vol. 192, pp. 107262, 2023.
[23] R. Reka, R. Karthick, R. S. Ram and G. Singh, “Multi head self-attention gated graph convolutional network based multi-attack intrusion detection in MANET,” Computers and Security, vol. 136, pp. 103526, 2024.
DOI:
https://doi.org/10.31449/inf.v50i2.13861Keywords:
business-finance-law-tax (BFLT), intelligent risk management, graph neural network, knowledge graph, multi-source heterogeneous dataDownloads
Published
Issue
Section
License
Authors retain copyright in their work. By submitting to and publishing with Informatica, authors grant the publisher (Slovene Society Informatika) the non-exclusive right to publish, reproduce, and distribute the article and to identify itself as the original publisher.
All articles are published under the Creative Commons Attribution license CC BY 3.0. Under this license, others may share and adapt the work for any purpose, provided appropriate credit is given and changes (if any) are indicated.
Authors may deposit and share the submitted version, accepted manuscript, and published version, provided the original publication in Informatica is properly cited.
