Building an e-learning system is a problem with multiple faces. By considering only computational preoccupations, one leads to a limited, rigid and closed system. We consider that problems emanating from social sciences, cognitive sciences and educational sciences are also important, especially to build the system's architecture and design and to evaluate and/or validate its performances. Our
... [Show full abstract] contribution proposes that the use of the models in the broad sense of the term makes it possible to limit the complexity of various e-Learning's spaces of problems to be specified at the time of building an e-Learning system (By the way, we have to define and discuss these spaces). These models are to be risen with definition of Interfaces and communication's flow between them. So that, the set of obtained specification, can be used to structure the e-Learning's system. In this way, each problem (production of contents, follow-up, definition of pedagogy...) will be associated a subset of models which after successive transformations will converge towards a data-processing solution.