In the context of distance learning and teaching, a re-engineering process needs a feedback on the learners' uses of the learning system. The feedback could be given by interviews, questionnaires, but in many systems, it is also provided by the analysis of log files. Our works on the re-engineering driven by models of e-learning systems, propose an architecture for usage analysis and a meta-language for describing the tracks? semantic and for linking them to observation needs defined in the predictive scenario. This architecture wants to be a collaborative cluster of services available for and designed by a community of practices on usage analysis. Concerning our meta-langage UTL, we consider that it is important to interpret tracks in order to compare the designer?s intentions with the learners? activities during a session.