The growing size and diversity of financial transaction data present significant challenges for modern auditing, particularly in achieving comprehensive audit coverage and reliable risk identification across regions. Traditional auditing methods, which rely on sampling or region-specific models, often result in limited coverage and inconsistent risk detection due to variations in data availability. To address these challenges, this paper proposes a big data-driven audit risk analysis framework that enables full transaction-level audit coverage. The framework incorporates a transfer learning-based risk scoring mechanism, allowing knowledge learned from data-rich regions to be effectively applied to data-scarce regions. This approach improves model robustness and supports scalable audit analysis across heterogeneous environments. The framework is evaluated using a simulated multi-regional setting, demonstrating its effectiveness in providing consistent and scalable audit risk insights. Overall, the proposed method enhances audit reliability and supports data-driven audit decision-making.
Xie et al. (Tue,) studied this question.