This thesis is concerned with the development of interactive systems for smart environments. One of the characteristic of smart environments is the need to support different interaction paradigms at runtime. On the one hand interaction is performed explicitly: the user performs an action in order to interact with the system (e.g., pressing a button to adjust the light). On the other hand actions of the user are interpreted by the smart environment, even though they have not been performed primarily to interact with the system: implicit interactions (e.g., walking to the speaker's desk to give a talk). A smart environment tries to infer those interactions to assist the user in her work (e.g., display slides at the projection canvas). Both interaction paradigms originate from different research fields and are currently treated independently although implicit and explicit interaction mutually influence each other and occur interleaved. The thesis introduces a task modeling language in order to cope with the given requirements of smart environments which can be used to perform interaction development for smart environments in a model-based fashion for both interaction paradigms.