Shevchenko O. Models and methods of cyber-social personnel management computing for critical systems

A Thesis is focused on reducing economic, technological and social expenses associated with minimizing failures in critical systems through increasing employee competencies and consistent excluding human from decision-making processes and replacing it by deterministic computing mechanisms using digital intelligent management based on metric monitoring of cyber-social processes and phenomena. The following results which characterised by scientific novelty are obtained: a structural model of computing, which is characterized by interactive online interaction between human, critical system and mechanisms of accurate digital monitoring-management, which makes it possible to exclude failures that lead to man-made disasters and social conflicts; a set-theoretical method of data retrieval, which is characterized by or-metrics for defining the similarities-differences of text fragments-objects, which allows determining the similarity of objects, the strategy of transforming one object into another, and identify the level of digital common interests, plagiarism, conflicts; a metric method of selecting employees for the given reference competencies that differs from analogues taking into account the background, psychophysiology, achievements, knowledge, skills, abilities, and makes it possible to reduce incorrect appointments to functional positions; a method for metric integrated evaluation of project (company) staff, which differs from analogues by searching for quasi-optimal coverage of job functionalities and allows to significantly reduce the time and material costs of the project; a computing method for decision-making that differs from analogues by exhaustive online monitoring and digital management and allows essential reducing operator errors when functioning of critical system; methods for segmental searching a subset of effective design solutions, which take into account the parameters of complexity and quality and differ from the existing analogs by structural differences between the sets of convex and nonconvex admissible variants. This results significantly reduce the time of analysis of input information for decision-making without reducing their quality.