Classification of Partitioning Problem Solution Methods on the bases of the Cluster Analysis

One of the most important stage during the computer-aided engineering is a stage of commutation circuit partitioning. It requires of designer to make decisions in weakly formalized domains. During the designing a decision maker faces with a lot of problems, one of which is a choice of partitioning problem solution method, which provides an effective solution. This problem belongs to NP-hard and NP-full problem. The goal of this work is partitioning problem solution methods classification against the performance criterion. In this article a classifier of partitioning problem solution methods on the bases of the cluster analysis was developed. A theoretical part of investigation observes cluster-analysis problem formalization by the example of partitioning problem solution methods, the main definitions are performed. A classifier algorithm on the bases of the cluster analysis is developed. A density based spatial clustering of applications with noise (DBSCAN) algorithm was chosen as the most effective for the problem solving. Domain ontologies were developed and visualized in Protege4.2. A suggested approach provides an opportunity to reduce time on qualitative solution generating in the performance of a classification problem.