Abstract. Validation is an important field in the software development process. It helps to increase the software quality but is also very expensive and time consuming. To decrease the costs approaches to automate the validation process are necessary. In this thesis a framework is developed, which does not need user interaction to validate the IBM InfoSphere Optim Performance Manager semantically. It is able to validate values of different behavioral patterns. It covers deterministic, semi-deterministic and non-deterministic behavior. The thesis describes the process of the development of the framework. It introduces available approaches and examines them with regard to the suitability for the framework. The found solution is described in theory and a prototype is implemented to apply the solution to praxis. This prototype is evaluated on the latest version of the IBM InfoSphere Optim Performance Manager.