A Quantitative Data Risk Assessment Model Based on Improved AHP

With the widespread adoption of big data technologies, data has emerged as a core factor of production; however, the associated risks of privacy leakage are becoming increasingly severe. Accurate and real-time assessment of data security risks is the prerequisite to implement effective privacy protection. Traditional data risk assessment methods, such as the Analytic Hierarchy Process (AHP), primarily rely on expert experience to construct judgment matrices. Consequently, these methods suffer from high subjectivity, low evaluation efficiency, and an inability to adapt to massive, dynamic data streams. To address these challenges, this paper proposes a dynamic data security risk assessment model based on an improved AHP driven by objective indicators. The improvements focus on three aspects: (1) establishing a six-dimensional evaluation system covering data sensitivity and distribution features; (2) replacing the subjective scoring at the alternative layer with quantitative mappings based on Tsallis entropy and logarithmic transformations; and (3) implementing a threshold-based absolute classification mechanism to supersede traditional relative ranking, thereby achieving automated decision-making. Furthermore, a Privacy Classification Protection (PCP) Platform is designed and implemented based on this model. Application verification on a realworld fitness dataset demonstrates that the model can accurately identify high-risk data features (e.g., abnormal distributions and high re-identification risks) and automatically output reasonable risk levels, significantly enhancing the efficiency and objectivity of data governance.