Creation and execution of semantic mappings between two (or more) ontologies is a core issue to enable interoperability across various applications in the Semantic Web. To handle the increasing number of individual ontologies, but also for being able to create mappings on the fly, it becomes necessary to develop automatic approaches. In this paper, we determine mappings based on the similarity of the features of individual ontological entities. We show that mappings can be derived automatically by encoding similarities into logical axioms. Processing these axioms by inference engines allows for detection, creation and processing of mappings on the fly without human intervention. The advantages of this approach are obvious. Firstly, the axioms can easily be reused for mappings of arbitrary ontologies, no additional modelling effort is required. Secondly, the inference engine is the only mandatory technological infrastructure which means that no additional implementation effort is needed. Finally, we evaluate our approach with very promising results.