Визначення методу кластеризації для вирішення задачі ранньої ідентифікації конфігураційних елементів ІТ-продукта

This study investigates the process that controls the IT project configuration. The task addressed is early identification of configuration items (CIs) within an enterprise management information system (IS). Research in this area is aimed at solving the task of early identification of services when refactoring software systems. Up to now, the application of artificial intelligence methods to define CIs has not been studied in detail. To solve the task of early identification of IS CIs, the density-based clustering algorithm Density-Based Spatial Clustering of Applications with Noise (DBSCAN) was adapted. The adapted DBSCAN was used to early define CIs in the functional task “Formation and maintenance of an individual plan of a scientific and pedagogical employee at a department”. As initial CIs, 10 functions and 12 entities in the database of the task were considered. The result is the sets of clusters that describe monolithic, modular, and service-oriented IS architectures. A comparative analysis of the use of DBSCAN, Divisive Analysis, Agglomerative Nesting, Chameleon, and k-means methods and algorithms for solving the task of early identification was carried out. The criteria “Cumbersome solution” and “Identification of separated CIs” were used for comparison. The application of DBSCAN made it possible to form a solution from one (monolithic and modular architecture) or two (service-oriented architecture) clusters and to detect separated CIs. These values of the proposed criteria are the best for the selected group of clustering methods and algorithms. Implementing the results makes it possible to automate a procedure for synthesizing the description of IS architecture. This automation would improve the quality of IS development by identifying a set of architectural entities of this IS for its design. This set is much smaller than the set of elementary IS functions