Nowadays, technology-enhanced learning systems must have the ability to reuse learning resources from distributed repositories, to take into account the context and to allow dynamic adaptation to different learners (individuals, dyad or team) based on substantial advances in pedagogical theories and knowledge models. We focus on learning systems using a problem-based learning approach represented by scenarios. In our framework, the goal of scenarios is to describe the learning and tutoring activities to acquire some knowledge domain (for instance physics) and know-how to solve a particular problem. The main issue is to design a generic scenario which can deal with most of learning situation for problem-based learning science curriculum. From a generic scenario, the learning system will compute on the fly a particular scenario dedicated to the current learner and its learning situation. The main contribution of this paper is a semantic and didactic-based model of scenarios for designing an adaptive and context-aware learning System. The scenario model is acquired from: i) the know-how and real practices of teachers ii) the theory in didactic anthropology of knowledge of Chevallard [1]; iii) a hierarchical task model.