Graph-based two-level indicator system construction method for smart city information security risk assessment

The rapid development of urban informatization has led to a deep integration of advanced information technology into urban life. Many decision-makers are starting to alleviate the adverse effects of this informatization process through risk assessment. However, existing methods cannot effectively analyze internal and hierarchical relationships because of the excessive number of indicators. Thus, it is necessary to construct an indicator's dependency graph and conduct a comprehensive hierarchical analysis to solve this problem. In this study, we proposed a graph-based two-level indicator system construction method. First, a random forest was used to extract the indicators' dependency graph from missing data. Then, spectral clustering was used to separate the graph and form a functional subgraph. Finally, PageRank was used to calculate the prioritization for each subgraph's indicator, and the two-level indicator system was established. To verify the performance, we took China's 25 smart cities as examples. For the simulation of risk level prediction, we compared our method with some machine learning algorithms, such as ridge regression, Lasso regression, support vector regression, decision trees, and multi-layer perceptron. Results showed that the two-level indicator system is superior to the general indicator system for risk assessment.