Abstract: In order to help person understand complex and large-scale ontologies, we presented a concept ranking algorithm for OWL (Web Ontology Language) ontologies. This algorithm is based on the ontological structure and semantic meanings of concepts/relations. Different from the traditional ranking methods, this algorithm applies semantic meanings (patterns) on concepts and relations. The importance of concepts reinforces one another in an iterative manner; only the semantically correct paths can flood relevant components of importance vector from a concept to its neighbors in ontology graphs. The experimental results show that such a semantically ranking method provides more reasonable ranking result than PageRank-like algorithms.