In the frame of the VOTECH project, we developed a prototype ontology of astronomical object types (e.g. Galaxy, Star, Emission-line star, X-ray source...) We present an automated process to create and test instances of the ontology's concepts from entries of astronomical objects data sources. The data associated with an entry such as its object types or its observational data (radial velocity, dimensions...) are translated to formal ontological concepts or properties, leading to the creation of an ontological instance corresponding to the entry. This instance can then be tested for consistency or used as an entry point to the ontology to infer implicit knowledge on the instantiated entry. Such an automated process could be used to either build persistent instance bases of astronomical objects or to create instances on-the-fly, the goal in both cases being to provide semantic inference capabilities to astronomical objects databases or repositories.