基于矩阵分析的车险欺诈团伙识别研究

iming at the dangerous collusion fraud in the vehicle insurance industry, this paper proposes a recognition algorithm based onn matrix analysis for auto insurance fraud gangs. The algorithm firstly starts from the relationship between the persons in the vehicle insurance cases, and builds a distribution matrix of people involved in the cases according to the relationship between the personnel and the case. Then the distance between the person vectors is calculated to construct the distance matrix, and finally the maximum is utillized. The principle of simillarity priority is used to cluser people. The experiment result shows that the method can reflect the degree of similarity between the characters accurately, and the same type of members after clustering also have greater relevance.