MODELING OF SOCIAL INTERACTIONS BASED ON GRAPH APPROACHES

Abstract

The article proposes an approach to modeling social interactions in organizational systems, which
consists of several stages: obtaining data about system users, for example, using network parsing; forming
a GH-model of the system based on fuzzy graphs with different types of vertices and multiple different
types of edges; calculating graph characteristics taking into account a certain type of edges; using values
of graph characteristics to analyze the system taking into account the inherent semantic load. The expediency
of using the GH graph for the study of social relations in organizational systems is substantiated,
since it has a number of advantages. The GH-graph allows you to set all the necessary multi-type relationships
and at the same time reduce the time of system analysis by 1.9 times by using multiple edges in the
form of a vector, allowing you to combine several different types of edges. Modification of the model consists
in using different types of vertices. The type of vertices in the graph is determined by calculating their
characteristics. The paper shows the process of forming a graph model of a subsystem and calculating its
characteristics. The results of calculating the degrees of vertices and their centrality by degrees are
shown. To calculate the metric characteristics of the graph model, a modified algorithm for finding shortest
paths in the GH-graph was used, which was previously developed. A special feature of this algorithm
is the ability to use filters based on the type of vertices and edges. Numerical indices of the radius and
diameter of the graph are obtained, groups of central and peripheral vertices are determined, the centrality
of vertices in proximity is calculated, taking into account the selected types of edges for the study of
different types of relations in the system. The analysis of the subsystem is carried out using the example of
solving two practical problems. Groups of employees of the enterprise were identified among the network
users, their possible statuses and communicative activities were determined. The user status refers to belonging
to groups of managers of different levels, a group of ordinary employees of the enterprise. A solution
to the problem of identifying users (groups of users) most suitable for the dissemination (or, conversely,
non-proliferation) of information on the network is proposed