Assessment of personal data safety parameters in power social networks based on topology

Abstract

The article investigates the model of the power social network, which, in contrast to classical approaches, allows to analyze the dynamic processes of interaction of individual agents within the network, in particular the dissemination of information about social impact. Expansion of social networks, connection of new nodes leads to an increase in the load on the system as a whole and negatively affects the protection of users, including their personal data. Traditionally, security parameters in social networks are studied using statistical methods and generalized mathematical dependencies and are fragmentary. The aim of the article is to develop a methodology for assessing the security parameters of personal data for networks with a degree distribution of connectivity of nodes based on the study of their topological features. The research is carried out on the basis of the classical Barabashi-Albert model using the principle of preferential connection, which puts the probability of making new connections depending on the number of existing connections of the node. A larger node means more opportunities to pick up new connections added to the network. The main security parameters are: the degree of the node, the average path length, the probability of joining new nodes, the clustering factor, the correlation between the degrees of neighboring nodes. It is shown that increasing the degree of the node and the length of the middle path has a negative effect on the protection of personal data, as it increases the likelihood of interception of information. Also, with increasing clustering factor, the flow of information increases, which leads to an increase in the load on the protection system and negatively affects the protection. Correlation between the degrees of neighboring nodes affects the redistribution of information flows and can, depending on the degree of nodes, both negatively and positively affect the protection. Modeling for networks of different scales is carried out and conclusions on expediency of application of a technique are made.