This paper is devoted to solving the problem of clustering OWL-objects. The popularity and the relevance of this problem are demonstrated in the article. The solution to the problem is also suitable for tasks such as semantic repository scaling, semantic data visualization and extraction of links between OWL-objects. Existing solutions use links between OWL-objects, but there are several limitations. Therefore, these solutions are not universal. Proposed approach overcomes some of these limitations. Model of S-tag has been introduced in this paper. S-tag allows one to use any thesaurus to add semantics into the classic clustering process. The S-tag clustering algorithm has been introduced here with a demonstration model features. A natural language sentence is the implementation of the S-tag model. Sentences of the OWL-object description can be used as S-tags attached to the OWL-objects. OWL-objects can be distributed across clusters of their S-tags. This approach is suitable when the separation of objects depends on the semantics of OWL-object descriptions. A set of OWL-objects with a small number of links has been clustered as an example of this approach. The clustering of this set makes finding of new links between OWL-objects possible. This set cant be clustered by existing solutions which use links between OWL-objects. It is planned to implement of S- tags and search engine indexing. Algorithms for the qualitative detection and extraction of tags in the text have to be also developed. In addition, there are plans for further studying of the properties of the S-tag model and the possibility of using this model in other areas.