The evolution of tactile devices allows the creation of applications supporting space manipulationwith more than 3 degrees of freedom. Often, each application requires a new grammar of gestures. The user needs to learn some new gestures often for an occasional use of application. This constraint is often blocking and users rapidly abandon application (to complex, to long). These grammars are inappropriate for occasional use (having information about a city, a movie, searching a restaurant) or for fun applications that notneedsa new long learning (educative games for example or fun games). A user would prefer to use a limited number of gestures. Our work is in the field of 3D geometry learning. This paper trying to prove that it's possibleto reduce set of multi-touch gestures that, developed from three basic actions (tap, pinch/spread, and drag). This work leverages the context to frame the gestures semantics so as to reduce the cognitive load of anybody that learning and using multi-touch gestures.