Many research efforts have led to robust database initial design methodologies. However, database evolutions suffer from lack of structured approaches. Using high-level conceptualization models like ontologies, we describe an original approach to database evolution management. Our aim is to help the maintenance engineer in 1) characterizing the evolution, 2) finding the best techniques and tools to deal with this evolution. Our guidance approach provides this help by linking formally a change ontology and a maintenance method ontology through an algorithmic approach. A historization process of evolutions enriches the set of information delivered to the maintenance engineer. Our future work consists in prototyping this approach.