One of the main functions of an information security system is the identification of any access subject to be able to investigate information security incidents. During executing procedures of scanning and vulnerability exploitation, qualified adversaries regularly change identifying features. Such operations can not only obfuscate logging the data in subsystems, thus, complicating the restoring of events chronology for an information security expert but also call into question the irrefutability of the evidence of participation of particular adversary to particular illegal operations. In the paper analyses of application of modern approaches of adversary identification in web resources, which does not require authentification of main part of users, is given (fingerprinting, analysis of behavioral features). Along with widely used in web analytics ''thermal maps'', user adapted profile and computer model of dynamics of ''user-mouse'' system, authors offer to identify the subjects of information security incident in readily available informational resources of the Internet. The main idea of the prospective approach consists of the following: when a thermal map is built, not only the density of data layout should be considered but also statistical parameters should be defined by an expert (the distance of intensity gradient, distance overlap, etc.). The authors also offer to consider the dynamics of user operations (e.g. calculation of the average duration of data entry into interactive elements). A description of each step of an appropriate technique and also information on its practical implementation are given. Robustness of the given approach is confirmed by a practical experiment. The offered technique is not a universal instrument of adversary identification. Only manual targeted attacks are considered, the cURL tools etc. used by adversaries are not taken into account. Therefore, it is recommended to use this technique exclusively in addition to working protective systems (WAF, IPS, IDS).