The article presents a solution to the problem of synthesizing training tasks and data arrays when organizing a computer training platform for conducting practical exercises to investigate information security incidents. Two main stages of the synthesis of the background and situational components of the conditionally real data arrays based on the structural-parametric model of interaction between users of information and telecommunication services are proposed: the formation of static and dynamic components. Static components are synthesized based on the method of forming the structure of social graphs using a composition of models for constructing complex networks with various structural parameters: the Watts-Strogatz model is used for a mobile communication service, and the Barabashi-Albert model is used for a social network service. To preserve the relationship between users in various services, a method is proposed for determining the largest common part of social graphs, based on the mutual differentiation of vertices and the allocation of a partial isomorphism of the compared graphs. When generating vertex attributes, the method of searching for social groups (families) is used, based on the Bron-Kerbosch algorithm for finding a click of a given size in a graph. For the synthesis of dynamic components of data arrays that describe the performance of communication events, the mathematical apparatus of color Petri nets is used. An interaction event in information and telecommunication services is represented in the form of a Petri net label, which contains the necessary set of parameters, depending on the type of service. It is proposed to use the structural, event, social and temporal statistical characteristics of real information and telecommunication services to form the initial marking of the Petri net.