Knowledge-based ontology concept for numerical data clustering

Abstract

Classical clustering algorithms are sufficiently well studied, they are used for grouping numerical data in similar structures - clusters. Similar objects are placed in the same cluster, different objects in another cluster. All of the classic clustering algorithms have common parameters, and successful selection of which also determines the clustering result. The most important parameters characterizing clustering are: clus- tering algorithm, metrics, initial number of clusters, criteria for clustering accuracy. In recent years, there has been a tendency towards the possibility of obtaining rules from clusters. Classical clustering algorithms do not apply semantic knowledge. It creates difficulties in interpreting the results of clustering. Presently, the use of ontology opportunities is developing very rapidly, that allows to gain knowledge about a certain data model. The paper analyzes the concept of ontology and prototype development for numerical data clusterization, which includes the most significant indicators characterizing clusterization. The aim of the work is to develop a concept for analyzing clustering data with the help of ontologies. As a result of the work, a study has been conducted on the use of ontologies in this type of tasks.