Dynamics of link formation in networks structured on the basis of predictive terms

Objectives. In order to model and analyze the information conductivity of complex networks having an irregular structure, it is possible to use percolation theory methods known in solid-state physics to quantify how close the given network is to a percolation transition, and thus to form a prediction model. Thus, the object of the study comprises international information networks structured on the basis of dictionaries of model predictive terms thematically related to cutting-edge information technologies.Methods. An algorithmic approach is applied to establish the sequence of combining the necessary operations for automated processing of textual information by the internal algorithms of specialized databases, software environments and shells providing for their integration during data transmission. This approach comprises the stages of constructing a terminological model of the subject area in the Scopus bibliographic database, then processing texts in natural language with the output of a visual map of the scientific landscape of the subject area in the VOSviewer program, and then collecting the extended data of parameters characterizing the dynamics of the formation of links of the scientific terminological network in the Pajek software environment.Results. Visual cluster analysis of the range of 645-3364 terms in the 2004-2021 dynamics of the memory and data storage technologies category, which are integrated into a total of 23 clusters, revealed active cluster formation in the field of the term quantum memory. On this basis, allowing qualitative conclusions are drawn concerning the local dynamics of the scientific landscape. The exploratory data analysis carried out in the STATISTICA software package indicates the correlation of the behavior of the introduced MADSTA keyword integrator with basic terms including periods of extremes, confirming the correctness of the choice of the methodology for detailing the study by year.Conclusions. A basis is established for the formation of a set of basic parameters required for an extensive computational modeling of a cluster formation in the semantic field of the scientific texts, especially in relation to simulations of the formation of the largest component of the network and percolation transitions.