with the wide application of information systems, more and more enterprises adopt ontology as conceptual backbone for business data management to take advantage of knowledge base and semantic web technology. However, dynamic business motivates modifying existed ontology according to the continuous changes, which is hard to be done manually due to the growing size of ontology. Thus automatic method is needed for ontology evolution. In this paper, an instance-driven ontology evolution approach is proposed to cope with the dynamic changes in ontology evolution. The proposed approach suggests changing directions to users to control the process of evolution. Then the changes of ontology are formalized with elementary and composite changes which are considered as the basis of evolution. Moreover, the identification of changes is realized by instances analyzing. Finally, the approach is tested in a hospital database. The result shows instance-driven ontology evolution is an efficient method to cope with data conception changes in enterprise data management.