Data-Driven Audit Risk Analytics for Balance-Sheet Recognition of Data Assets under Blockchain-Enabled Evidence Governance
Main article
Abstract
The balance-sheet recognition of data assets has become a critical business analytics problem because firms, auditors, and regulators must convert fluid digital resources into verifiable accounting objects without allowing strategic overvaluation, weak evidence, or collusive assurance to distort market signals. Existing studies have mainly examined data asset governance through institutional interpretation, blockchain-enabled traceability, or evolutionary games among firms, auditors, and regulators. This article develops a related but distinct data-driven framework that treats data asset recognition as an audit risk analytics task. We construct a cost-sensitive predictive model that integrates evidence-quality indicators, valuation-dispersion variables, auditor-independence signals, firm-control histories, and blockchain evidence-coverage measures. A calibrated numerical panel of 1,200 data asset recognition applications across six digital-economy sectors is used to compare logistic regression, classification trees, random forest, gradient boosting, extreme gradient boosting, and support vector machines. The best-performing model reaches an AUC of 0.907, a PR-AUC of 0.661, and a cost-weighted F1 score of 0.681, while reducing expected misrecognition loss by 34.7% relative to rule-based screening. Scenario analysis indicates that blockchain evidence governance improves compliance most strongly when it is coupled with cost-sensitive analytics and targeted audit escalation rather than applied as a stand-alone technical ledger. Sensitivity analysis further shows that evidence coverage, valuation dispersion, and auditor-client repetition jointly determine whether regulatory rewards and penalties produce stable compliance. The study contributes to business and data analytics by transforming the governance of data assets from a purely normative or game-theoretic issue into an operational risk-scoring, audit-prioritization, and decision-optimization problem.
