Abstract: Thefts， robberies and drug cases from 2012 to 2017 in Chicago，United States are selected to study. A method of constructing criminal network based on spatio-temporal influence domain is proposed， and the mapping analysis is carried out according to the behavior characteristics of criminal manifestation and network behavior characteristics. The experimental results show that the criminal networks constructed by the proposed method exhibit scale-free and small-world characteristics under different crime types and different network scales. The local area showed high aggregation and different aggregation characteristics. The degree of corporatization of criminal networks is obvious. It can be seen that complex network means can be used to conduct further in-depth discussion on criminal network modeling and topological characteristics analysis， so as to provide new ideas for the dynamics study of different criminal patterns， prevention and resource allocation.

Key words: criminal network; network behavior characteristics; topological characteristics; central index