Development of Impact Factors Reverse Analysis Method for Software Complexes’ Support Automation

This research represents a corresponding and developed specialized impact factors reverse analysis method for software complexes’ support automation; it is intended for the analysis of impact factors affecting the supported software’s (or processes of its comprehensive support) subjective perception results, as one of the constituent tasks of the more complex problem of software complexes’ support automation. The developed method provides the possibility to restore certain boundaries of impact factors by classifying the multilayer perceptron’s hidden layer neurons and calculating the probability coefficients of the belonging of these neurons to the corresponding specific pre-determined impact factors. The problem of determining the influence of impact factors on the subjective perception of the object of support (the supported software or the processes of its comprehensive support) by the relevant subjects (interacting with this object, providing and implementing its support) was resolved through the approach developed and proposed by the authors in the scope of this research. A key feature of the proposed approach is to assign the neurons of the hidden layers (of the multilayer perceptron type of artificial neural networks) functional–semantic meaning(s), which they have been deprived of a priori, performing (before this) by default an exclusively operational (calculation) function mainly for the correctness of the training and functioning of the multilayer perceptron itself. The potential of the developed method allows us to apply it for solving a huge number of applied practical tasks, such as the one provided in the scope of this research, which is as follows: a practical task of the support team members’ portrait determination, followed by a further search (detection) of the interchangeable members of this support team to ensure the possibility of quick transfer of the stack of tickets (which are in the middle of the active resolution process) between these members.