Abstract: In order to address the lack of scientific quantitative evaluation on passenger security screening classification of civil aviation in the future, an evaluation system was established by using AHP to realize passenger background analysis and risk warning. Considering the shortcoming of traditional AHP method, namely its high subjectivity as evaluation indicators were selected through artificial experience, an improved AHP was proposed. A sample database describing personal characteristics was created through JJATT, and RF was applied to screen characteristics of the database before a personnel risk evaluation system was developed with optimal combination of characteristics as evaluation indicators. Then, index weights were determined through AHP method, and a risk assessment method for key personnel was put forward based on evaluation index scoring table. Finally, the method was applied to evaluate risk levels of personnel in large sample cases and real cases. The results show that the evaluation accuracy rate of proposed method for multiple sample sizes is above 92%, and it can evaluate personnel risk level of samples. And improved AHP, as the core of the proposed method, is verified to have certain rationality.

Key words: analytic hierarchy process (AHP), key personnel, risk evaluation, John Jay & ARTIS transnational terrorism database (JJATT), random forest (RF)