The paper analyzes approaches to the design of ontological models of adaptive software systems. To evaluate the quality of the projected ontologies, specialized groups of metrics are used, which allow us to determine the shortcomings in the structure and functioning of the model. An analysis of the use of ontological models in the process of designing and developing adaptive software systems has been carried out. It was established that the ontological approach allows for a more effective presentation of the subject area, which improves the software configuration process. The use of ontological models allows flexible responses to changes in user requirements without the need to directly update the source code of the software. The process of designing ontologies based on classical and abstract approaches is presented. It was determined that the use of a classical approach to the design of ontological models could complicate the adaptation process due to the need to change the structure and reconfiguration of the system. To solve this problem, an abstract approach is proposed that ensures the dynamism of the adaptation process, without the need for a complete reconfiguration of the system. A set of specialized metrics is defined that allows for the analysis of the ontological model of the software system, the evaluation of its syntactic and semantic quality, compliance with rules and interoperability, and the possibility of expansion. A comparative analysis of abstract and classical approaches to the design of ontological models was carried out based on defined structural metrics, as well as metrics of the scheme and filling of the knowledge base. The obtained values of structural and hierarchical metrics confirm the quality and effectiveness of the proposed approaches. According to the defined metrics, the absence of cycles and entanglement in the ontological model indicates the correctness of the construction of the structure from the point of view of ergonomics and the possibility of further changing the structure and filling of the ontological model. The analyzed values of the schema metrics and the filling of the knowledge base demonstrate the better effectiveness of the abstract approach in the process of determining the software configuration, which provides faster and more correct processing of ontological SWRL rules during adaptation.