Abstract: for the modeling of QoS-aware semantic Web service, an upper-level ontology, named as WS-QMO, is proposed to describe and publish both QoS information and QoS requirement in a semantic way. By using WS-QMO, QoS attributes and metrics are defined by OWL concepts and roles, and the QoS constraints are defined by SWRL rules. As far as we know, the WS-QMO ontology satisfies most of the current requirements for QoS-based Web service description. In addition, due to the proper integration of SWRL rules, the syntax of WS-QMO is rather concise, and it has richer description capability on complex QoS attributes comparing to the existing Web service modeling ontology. Furthermore, WS-QMO based QoS knowledge base is used in our semantic Web service architecture, which has dynamic description logic as its logic foundation. In this knowledge base, the constraint verification problem can be solved as a reasoning problem of the description logic with SWRL rules. The main KB management mechanism is built on top of the algorithms for constraint verification. More important, this mechanism comes up with a new automated Web service discovery algorithm that supports complex QoS constraints. The discovery algorithm makes a balance between semantic description capability and computational complexity.