A Dynamic Risk Assessment Model for Cross-Border E-Commerce Data Integrating Genetic Algorithms and Zero-Knowledge Proofs

Abstract

This paper proposes a dynamic risk assessment model for cross-border e-commerce data that integrates genetic algorithms with zero-knowledge proofs. The study first constructs a dynamic evaluation framework based on multidimensional feature extraction and employs an improved genetic algorithm featuring an adaptive crossover and mutation operator to efficiently optimize risk assessment parameters. Subsequently, a lightweight zero-knowledge proof protocol is designed to verify data integrity and the trustworthiness of the risk evaluation logic, while ensuring the confidentiality of the original data. Experimental results on a simulated cross-border e-commerce dataset demonstrate that the proposed model achieves 92.7% accuracy on a simulated dataset, surpassing traditional SVM and Random Forest by 15.3% and 7.5%, respectively. Additionally, the proof generation time is maintained at the millisecond level, significantly outperforming existing homomorphic encryption schemes. This research offers an innovative solution to the “security-efficiency-compliance” trilemma in cross-border e-commerce data flows and holds substantial practical significance for constructing a trusted data element circulation system.