Scientific Journal of KubSAU

The article deals with the problem of increasing the efficiency of the control and prevention of leakage of confidential information circulating within the organization, and in cooperation with the information networks of general use through the development of intelligent system of pragmatic identification of confidential information. The main purpose of this system is to develop early detection, prevention and localization (minimization) of the risks associated with violation of the order of processing, storage and transmission of confidential information. The article presents an approach to the organization of the knowledge base of intelligent system of pragmatic identification of information of a confidential nature, using two models of knowledge representation. We have defined a subject field to meet the challenges of which the intelligent system was meant for. The article describes the ontological model of knowledge representation used for formal representation of domain concepts in the knowledge base. It presents an indicator of exactness of meaning reference to a concept of domain meaning. The article describes a confidence index for each rule from the rule base that characterizes the measure of the likelihood of a conclusion when a rule implemented. It presents a general structure of the rules base, conditions of execution of each rule of conclusion. To accelerate the work of intelligent system a parallel search for the rules in the knowledge base is to be applied. In conclusion, the algorithm of parallel search is described

In the article, we develop the software for process of confidential data detection based on the knowledge bases technology. The algorithm of detection of confidential data in the electronic documents transferred outbound of protected information system, due to application of linguistic technologies of the thorough text analysis is presented. The estimation of computing complexity of the developed algorithm is made up. The developed data structures and algorithm are realized in the programming language C++. Results of the experiments, confirmed workability of the developed algorithm are presented. The performed experimental researches of the developed algorithm have shown its practical applicability. The estimation of quality of confidential data detection is made up. The obtained estimations of detection quality have shown, that the developed structures of data and algorithm provide more effective and qualitative solution of problems of the confidential information detection in electronic documents, at due to application of knowledge bases technology where the subject domain of the analyzed information is considered. Distinctive feature of the developed algorithm of confidential data detection is the use of rapid access to hashed concept ontographs simultaneous implementation of knowledge base rules, which allows raising indicators of completeness and accuracy of detection. A scope of application of the developed software is the protection frames of the information intended for revealing of an information transfer containing data of confidential character in electronic documents outbound of protected information system with violation of security policy