This paper deals with the subject of personalisation in Technology-Enhanced Learning (TEL) Systems. As we want user centred solutions, we claim that considerable advantage can be gained from allowing observation and interpretation of learners’ interactions traces in TEL systems. The key concepts of our approach are modelled traces, and trace transformations. These transformations such as abstraction or filtering enable a better understanding of the interactions taking place in the TEL systems and allow to provide services for personalisation.