In the last decades, software development turned towards flexibly combinable software services. Among the services provided in world-wide markets, the services that match a requesters functional and non-functional requirements best need to be discovered by comparing the specifications of the requirements and the services. However, existing matching approaches only consider a specific kind of requirements and the more complex the specifications, the more the probability for imperfections increases. While requesters often have vague requirements, providers provide incomplete specifications. In these cases, traditional matching approaches deliver adulterated results. In order to deal with complex service specifications, we propose the idea of matching processes that combine multiple existing matching approaches and aggregate their results. For this purpose, a model-driven framework called MatchBox, is introduced. In order to cope with uncertainty, we propose Fuzzy Matching. On the basis of well-defined fuzziness sources and types, the amount of induced fuzziness is quantified and returned as part of an informative matching result. This improves the decision-making of both service requesters and service providers. By combining multiple research areas in a novel way, this thesis describes concepts that go beyond the state of the art in service matching. Thereby, it constitutes an important step to bring service matching into practice.