Web Service is a major computing resource and a main software paradigm. With rapid increase of WebServices in recent years, methods of accurately discovering the required service are becoming a research focus. Byusing service matching method, based on conception similarity, in this paper, the study transforms user requirementsand Semantic Web Service description documents OWL-S profile into ontology trees separately. Next, byhierarchical and taxonomical method, according to the ontology trees, the study computes the conception similarity,attribute similarity, and structure similarity respectively, which effectively avoids complex reason. Thereafter,according to the relationship between conception similarity and structure similarity, the study defines the sets ofconstraints and restructures the requirement trees with constraints to improve the precision and recall of the servicediscovery. Finally, the paper presents the algorithm of Semantic Web Service discovery and conducts experimentsby developing the prototype system OWLS-CSR, which proves the feasibility and affectivity of this method.