In order to enable interoperability between ontology-based systems, ontology matching techniques have been proposed. However, when the generated mappings suffer from logical flaws, their usefulness may be diminished. In this paper we present an approximate method to detect and correct violations to the so-called conservativity principle where novel subsumption entailments between named concepts in one of the input ontologies are considered as unwanted. We show that this is indeed the case in our application domain based on the EU Optique project. Additionally, our extensive evaluation conducted with both the Optique use case and the data sets from the Ontology Alignment Evaluation Initiative (OAEI) suggests that our method is both useful and feasible in practice.