Search for functional-process gaps using a graph neural network in a holacratic organizational system.

The article is devoted to the problem of searching for functional-processual gaps in a holacratic organizational system. The purpose of the study. Graph neural networks are an applied tool for working with a large array of unstructured data, including those representing a description of organizational systems, allowing you to organize the connections of business processes and functions performed in the form of graphs. In turn, the connections formed between graphs allow you to solve the problem of finding functional gaps and anomalies in organizational systems of any type, including holacratic ones. Materials and methods. The paper proposes an approach to finding functional gaps in processes taking into account the branched management structure in a large organizational system. The approach bases on data obtained in the form of machine-readable text from regulations on departments, orders on the distribution of powers and job descriptions of a holacratic organization. The method is implemented through the construction of a tree of functions and the identification of functional-process gaps, including those along intersecting branches. Results. The result of the study is a mathematical model based on graph embeddings, presented in the form of text vectorization. The graph neural network developed on the basis of the mathematical model allows one to move on to managing a holacratic organizational system through a comparison of functions according to regulatory and administrative documentation to solve the problem of identifying functional and process gaps. Modeling of the functions of a holacratic organization is presented in the form of a tree, and the computational values obtained using the graph neural network allow one to identify functional and process gaps. Conclusion. The obtained results allow to initiate cycles of changing functions in a holacratic organizational system for the purpose of subsequent optimization of the cost of process execution. As a basic recommendation, it is proposed to use the developed graph neural network in holacratic organizational systems to search for functional-process gaps for the purpose of subsequent optimization of both the management model and individual processes.