The service-oriented architecture (SOA) proposes a new software development paradigm that is based on loosely coupled software components deployed and located across the web. This key benefit of SOA allows building workflows crossing organizational boundaries. One of the major challenges in this field of research is the automatic discovery and composition of services. In this paper we suggest a novel ontology-based discovery and composition approach. In a first step standard web services descriptions based on WSDL are semantically enriched by pre-processing steps including general-purpose ontologies as well as domainspecific ontologies. Resulting semantic-aware service profiles are stored in a registry component. In a second step we describe the matchmaking algorithms underlying the IRIS discovery component. Finally, we evaluate the performance of our algorithms analytically as well as we measure its quality by an experimental setup.